anki/Cargo.lock

5190 lines
122 KiB
Plaintext
Raw Normal View History

# This file is automatically @generated by Cargo.
# It is not intended for manual editing.
2021-06-25 07:35:25 +02:00
version = 3
[[package]]
name = "addr2line"
version = "0.19.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "a76fd60b23679b7d19bd066031410fb7e458ccc5e958eb5c325888ce4baedc97"
dependencies = [
"gimli",
]
[[package]]
name = "adler"
2021-03-07 10:04:34 +01:00
version = "1.0.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
2021-03-07 10:04:34 +01:00
checksum = "f26201604c87b1e01bd3d98f8d5d9a8fcbb815e8cedb41ffccbeb4bf593a35fe"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
[[package]]
name = "aes"
version = "0.7.5"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "9e8b47f52ea9bae42228d07ec09eb676433d7c4ed1ebdf0f1d1c29ed446f1ab8"
dependencies = [
"cfg-if",
2022-11-30 03:32:27 +01:00
"cipher 0.3.0",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
"cpufeatures",
"opaque-debug",
]
[[package]]
name = "ahash"
version = "0.7.6"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "fcb51a0695d8f838b1ee009b3fbf66bda078cd64590202a864a8f3e8c4315c47"
2021-06-25 08:22:21 +02:00
dependencies = [
"getrandom 0.2.8",
2021-06-25 08:22:21 +02:00
"once_cell",
"version_check",
]
[[package]]
name = "aho-corasick"
version = "0.7.20"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "cc936419f96fa211c1b9166887b38e5e40b19958e5b895be7c1f93adec7071ac"
dependencies = [
"memchr",
]
[[package]]
name = "ammonia"
version = "3.3.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "64e6d1c7838db705c9b756557ee27c384ce695a1c51a6fe528784cb1c6840170"
dependencies = [
"html5ever",
"maplit",
2022-09-24 04:39:21 +02:00
"once_cell",
"tendril",
"url",
]
2022-09-24 04:39:21 +02:00
[[package]]
name = "android_system_properties"
version = "0.1.5"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "819e7219dbd41043ac279b19830f2efc897156490d7fd6ea916720117ee66311"
dependencies = [
"libc",
]
[[package]]
name = "anes"
version = "0.1.6"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "4b46cbb362ab8752921c97e041f5e366ee6297bd428a31275b9fcf1e380f7299"
2020-11-24 09:41:03 +01:00
[[package]]
name = "anki"
version = "0.0.0"
dependencies = [
"ammonia",
"anki_i18n",
"anki_io",
"anki_proto",
"anyhow",
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
"async-compression",
"async-stream",
2021-01-05 10:58:53 +01:00
"async-trait",
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
"axum",
"axum-client-ip",
2020-11-24 09:41:03 +01:00
"blake3",
"bytes",
2020-11-24 09:41:03 +01:00
"chrono",
"coarsetime",
"convert_case",
"criterion",
2022-09-24 04:39:21 +02:00
"csv",
"difflib",
2020-11-24 09:41:03 +01:00
"env_logger",
"flate2",
"fluent",
2021-03-27 04:24:11 +01:00
"fluent-bundle",
"fnv",
2020-11-24 09:41:03 +01:00
"futures",
"hex",
"htmlescape",
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
"hyper",
V3 parent limits (#1638) * avoid repinning Rust deps by default * add id_tree dependency * Respect intermediate child limits in v3 * Test new behaviour of v3 counts * Rework v3 queue building to respect parent limits * Add missing did field to SQL query * Fix `LimitTreeMap::is_exhausted()` * Rework tree building logic https://github.com/ankitects/anki/pull/1638#discussion_r798328734 * Add timer for build_queues() * `is_exhausted()` -> `limit_reached()` * Move context and limits into `QueueBuilder` This allows for moving more logic into QueueBuilder, so less passing around of arguments. Unfortunately, some tests will require additional work to set up. * Fix stop condition in new_cards_by_position * Fix order gather order of new cards by deck * Add scheduler/queue/builder/burying.rs * Fix bad tree due to unsorted child decks * Fix comment * Fix `cap_new_to_review_rec()` * Add test for new card gathering * Always sort `child_decks()` * Fix deck removal in `cap_new_to_review_rec()` * Fix sibling ordering in new card gathering * Remove limits for deck total count with children * Add random gather order * Remove bad sibling order handling All routines ensure ascending order now. Also do some other minor refactoring. * Remove queue truncating All routines stop now as soon as the root limit is reached. * Move deck fetching into `QueueBuilder::new()` * Rework new card gather and sort options https://github.com/ankitects/anki/pull/1638#issuecomment-1032173013 * Disable new sort order choices ... depending on set gather order. * Use enum instead of numbers * Ensure valid sort order setting * Update new gather and sort order tooltips * Warn about random insertion order with v3 * Revert "Add timer for build_queues()" This reverts commit c9f5fc6ebe87953c17a0c842990b009b5596c69c. * Update rslib/src/storage/card/mod.rs (dae) * minor wording tweaks to the tooltips (dae) + move legacy strings to bottom + consistent capitalization (our leech action still needs fixing, but that will require introducing a new 'suspend card' string as the existing one is used elsewhere as well)
2022-02-10 00:55:43 +01:00
"id_tree",
"inflections",
2021-10-02 12:42:03 +02:00
"itertools",
2020-11-24 09:41:03 +01:00
"lazy_static",
"nom",
2020-11-24 09:41:03 +01:00
"num-integer",
Backups (#1685) * Add zstd dep * Implement backend backup with zstd * Implement backup thinning * Write backup meta * Use new file ending anki21b * Asynchronously backup on collection close in Rust * Revert "Add zstd dep" This reverts commit 3fcb2141d2be15f907269d13275c41971431385c. * Add zstd again * Take backup col path from col struct * Fix formatting * Implement backup restoring on backend * Normalize restored media file names * Refactor `extract_legacy_data()` A bit cumbersome due to borrowing rules. * Refactor * Make thinning calendar-based and gradual * Consider last kept backups of previous stages * Import full apkgs and colpkgs with backend * Expose new backup settings * Test `BackupThinner` and make it deterministic * Mark backup_path when closing optional * Delete leaky timer * Add progress updates for restoring media * Write restored collection to tempfile first * Do collection compression in the background thread This has us currently storing an uncompressed and compressed copy of the collection in memory (not ideal), but means the collection can be closed without waiting for compression to complete. On a large collection, this takes a close and reopen from about 0.55s to about 0.07s. The old backup code for comparison: about 0.35s for compression off, about 8.5s for zip compression. * Use multithreading in zstd compression On my system, this reduces the compression time of a large collection from about 0.55s to 0.08s. * Stream compressed collection data into zip file * Tweak backup explanation + Fix incorrect tab order for ignore accents option * Decouple restoring backup and full import In the first case, no profile is opened, unless the new collection succeeds to load. In the second case, either the old collection is reloaded or the new one is loaded. * Fix number gap in Progress message * Don't revert backup when media fails but report it * Tweak error flow * Remove native BackupLimits enum * Fix type annotation * Add thinning test for whole year * Satisfy linter * Await async backup to finish * Move restart disclaimer out of backup tab Should be visible regardless of the current tab. * Write restored collection in chunks * Refactor * Write media in chunks and refactor * Log error if removing file fails * join_backup_task -> await_backup_completion * Refactor backup.rs * Refactor backup meta and collection extraction * Fix wrong error being returned * Call sync_all() on new collection * Add ImportError * Store logger in Backend, instead of creating one on demand init_backend() accepts a Logger rather than a log file, to allow other callers to customize the logger if they wish. In the future we may want to explore using the tracing crate as an alternative; it's a bit more ergonomic, as a logger doesn't need to be passed around, and it plays more nicely with async code. * Sync file contents prior to rename; sync folder after rename. * Limit backup creation to once per 30 min * Use zstd::stream::copy_decode * Make importing abortable * Don't revert if backup media is aborted * Set throttle implicitly * Change force flag to minimum_backup_interval * Don't attempt to open folders on Windows * Join last backup thread before starting new one Also refactor. * Disable auto sync and backup when restoring again * Force backup on full download * Include the reason why a media file import failed, and the file path - Introduce a FileIoError that contains a string representation of the underlying I/O error, and an associated path. There are a few places in the code where we're currently manually including the filename in a custom error message, and this is a step towards a more consistent approach (but we may be better served with a more general approach in the future similar to Anyhow's .context()) - Move the error message into importing.ftl, as it's a bit neater when error messages live in the same file as the rest of the messages associated with some functionality. * Fix importing of media files * Minor wording tweaks * Save an allocation I18n strings with replacements are already strings, so we can skip the extra allocation. Not that it matters here at all. * Terminate import if file missing from archive If a third-party tool is creating invalid archives, the user should know about it. This should be rare, so I did not attempt to make it translatable. * Skip multithreaded compression on small collections Co-authored-by: Damien Elmes <gpg@ankiweb.net>
2022-03-07 06:11:31 +01:00
"num_cpus",
2020-11-24 09:41:03 +01:00
"num_enum",
"once_cell",
"percent-encoding-iri",
2021-03-07 10:04:34 +01:00
"pin-project",
2020-11-24 09:41:03 +01:00
"prost",
"prost-build",
"prost-reflect",
"prost-types",
2022-09-24 04:39:21 +02:00
"pulldown-cmark 0.9.2",
"rand 0.8.5",
2020-11-24 09:41:03 +01:00
"regex",
2021-10-02 12:42:03 +02:00
"reqwest",
2020-11-24 09:41:03 +01:00
"rusqlite",
"scopeguard",
"serde",
"serde-aux",
"serde_json",
"serde_repr",
"serde_tuple",
2022-11-30 03:32:27 +01:00
"sha1",
"snafu",
2021-03-07 10:08:03 +01:00
"strum",
2020-11-24 09:41:03 +01:00
"tempfile",
"tokio",
"tokio-util",
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
"tower-http",
"tracing",
"tracing-appender",
"tracing-subscriber",
"unic-ucd-category",
2020-11-24 09:41:03 +01:00
"unicase",
"unicode-normalization",
"utime",
"which",
"windows",
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
"wiremock",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
"workspace-hack",
2022-11-30 03:32:27 +01:00
"zip",
"zstd 0.12.3+zstd.1.5.2",
2020-11-24 09:41:03 +01:00
]
[[package]]
name = "anki_i18n"
version = "0.0.0"
dependencies = [
"anki_io",
"anyhow",
"fluent",
2021-03-27 04:24:11 +01:00
"fluent-bundle",
"fluent-syntax",
"inflections",
"intl-memoizer",
"num-format",
2022-09-24 04:39:21 +02:00
"phf 0.11.1",
"serde",
"serde_json",
"unic-langid",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
"workspace-hack",
]
[[package]]
name = "anki_i18n_helpers"
version = "0.0.0"
dependencies = [
"fluent-syntax",
"lazy_static",
"regex",
"serde_json",
"walkdir",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
"workspace-hack",
]
[[package]]
name = "anki_io"
version = "0.0.0"
dependencies = [
"snafu",
"tempfile",
]
[[package]]
name = "anki_proto"
version = "0.0.0"
dependencies = [
"anki_io",
"anyhow",
"inflections",
"num_enum",
"prost",
"prost-build",
"prost-reflect",
"prost-types",
"serde",
"snafu",
"strum",
]
[[package]]
name = "anstream"
version = "0.2.6"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "342258dd14006105c2b75ab1bd7543a03bdf0cfc94383303ac212a04939dff6f"
dependencies = [
"anstyle",
"anstyle-parse",
"anstyle-wincon",
"concolor-override",
"concolor-query",
"is-terminal",
"utf8parse",
]
[[package]]
name = "anstyle"
version = "0.3.5"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "23ea9e81bd02e310c216d080f6223c179012256e5151c41db88d12c88a1684d2"
[[package]]
name = "anstyle-parse"
version = "0.1.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "a7d1bb534e9efed14f3e5f44e7dd1a4f709384023a4165199a4241e18dff0116"
dependencies = [
"utf8parse",
]
[[package]]
name = "anstyle-wincon"
version = "0.2.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "c3127af6145b149f3287bb9a0d10ad9c5692dba8c53ad48285e5bec4063834fa"
dependencies = [
"anstyle",
"windows-sys 0.45.0",
]
[[package]]
name = "anyhow"
version = "1.0.71"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "9c7d0618f0e0b7e8ff11427422b64564d5fb0be1940354bfe2e0529b18a9d9b8"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
[[package]]
name = "apple-bundles"
version = "0.17.0"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "716b8a7bacf7325eb3e7a1a7f5ead4da91e1e16d9b56a25edea0e1e4ba21fd8e"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
dependencies = [
"anyhow",
"plist",
2022-11-30 03:32:27 +01:00
"simple-file-manifest",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
"walkdir",
]
[[package]]
name = "archives"
version = "0.0.0"
dependencies = [
"anki_io",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
"camino",
"flate2",
"reqwest",
2022-11-30 03:32:27 +01:00
"sha2",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
"tar",
"tokio",
"workspace-hack",
"xz2",
2022-11-30 03:32:27 +01:00
"zip",
"zstd 0.12.3+zstd.1.5.2",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
]
[[package]]
name = "arrayref"
version = "0.3.7"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "6b4930d2cb77ce62f89ee5d5289b4ac049559b1c45539271f5ed4fdc7db34545"
2021-10-02 12:42:03 +02:00
[[package]]
name = "arrayvec"
2021-11-18 11:54:00 +01:00
version = "0.7.2"
2021-10-02 12:42:03 +02:00
source = "registry+https://github.com/rust-lang/crates.io-index"
2021-11-18 11:54:00 +01:00
checksum = "8da52d66c7071e2e3fa2a1e5c6d088fec47b593032b254f5e980de8ea54454d6"
2021-10-02 12:42:03 +02:00
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
[[package]]
name = "assert-json-diff"
version = "2.0.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "47e4f2b81832e72834d7518d8487a0396a28cc408186a2e8854c0f98011faf12"
dependencies = [
"serde",
"serde_json",
]
[[package]]
name = "async-channel"
version = "1.8.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "cf46fee83e5ccffc220104713af3292ff9bc7c64c7de289f66dae8e38d826833"
dependencies = [
"concurrent-queue",
"event-listener",
"futures-core",
]
[[package]]
name = "async-compression"
version = "0.3.15"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "942c7cd7ae39e91bde4820d74132e9862e62c2f386c3aa90ccf55949f5bad63a"
dependencies = [
"futures-core",
"memchr",
"pin-project-lite",
"tokio",
"zstd 0.11.2+zstd.1.5.2",
"zstd-safe 5.0.2+zstd.1.5.2",
]
[[package]]
name = "async-stream"
version = "0.3.4"
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "ad445822218ce64be7a341abfb0b1ea43b5c23aa83902542a4542e78309d8e5e"
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
dependencies = [
"async-stream-impl",
"futures-core",
"pin-project-lite",
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
]
[[package]]
name = "async-stream-impl"
version = "0.3.4"
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "e4655ae1a7b0cdf149156f780c5bf3f1352bc53cbd9e0a361a7ef7b22947e965"
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
dependencies = [
"proc-macro2",
"quote",
"syn 1.0.109",
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
]
2021-01-05 10:58:53 +01:00
[[package]]
name = "async-trait"
version = "0.1.68"
2021-01-05 10:58:53 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "b9ccdd8f2a161be9bd5c023df56f1b2a0bd1d83872ae53b71a84a12c9bf6e842"
2021-01-05 10:58:53 +01:00
dependencies = [
"proc-macro2",
"quote",
"syn 2.0.12",
2021-01-05 10:58:53 +01:00
]
[[package]]
name = "atty"
version = "0.2.14"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "d9b39be18770d11421cdb1b9947a45dd3f37e93092cbf377614828a319d5fee8"
dependencies = [
"hermit-abi 0.1.19",
"libc",
"winapi",
]
[[package]]
name = "autocfg"
version = "1.1.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "d468802bab17cbc0cc575e9b053f41e72aa36bfa6b7f55e3529ffa43161b97fa"
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
[[package]]
name = "axum"
version = "0.6.12"
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "349f8ccfd9221ee7d1f3d4b33e1f8319b3a81ed8f61f2ea40b37b859794b4491"
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
dependencies = [
"async-trait",
"axum-core",
"axum-macros",
"bitflags 1.3.2",
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
"bytes",
"futures-util",
"headers",
"http",
"http-body",
"hyper",
"itoa",
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
"matchit",
"memchr",
"mime",
"multer",
"percent-encoding",
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
"pin-project-lite",
"rustversion",
"serde",
"serde_json",
"serde_path_to_error",
"serde_urlencoded",
"sync_wrapper",
"tokio",
"tower",
"tower-layer",
"tower-service",
]
[[package]]
name = "axum-client-ip"
version = "0.3.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "ddfb5a3ddd6367075d50629546fb46710584016ae7704cd03b6d41cb5be82e5a"
dependencies = [
"axum",
"forwarded-header-value",
]
[[package]]
name = "axum-core"
version = "0.3.3"
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "b2f958c80c248b34b9a877a643811be8dbca03ca5ba827f2b63baf3a81e5fc4e"
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
dependencies = [
"async-trait",
"bytes",
"futures-util",
"http",
"http-body",
"mime",
"rustversion",
"tower-layer",
"tower-service",
]
[[package]]
name = "axum-macros"
version = "0.3.7"
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "2bb524613be645939e280b7279f7b017f98cf7f5ef084ec374df373530e73277"
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
dependencies = [
"heck",
"proc-macro2",
"quote",
"syn 2.0.12",
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
]
[[package]]
name = "backtrace"
version = "0.3.67"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "233d376d6d185f2a3093e58f283f60f880315b6c60075b01f36b3b85154564ca"
dependencies = [
"addr2line",
"cc",
"cfg-if",
"libc",
"miniz_oxide",
"object",
"rustc-demangle",
]
[[package]]
name = "base64"
version = "0.13.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "9e1b586273c5702936fe7b7d6896644d8be71e6314cfe09d3167c95f712589e8"
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
[[package]]
name = "base64"
version = "0.21.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "a4a4ddaa51a5bc52a6948f74c06d20aaaddb71924eab79b8c97a8c556e942d6a"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
[[package]]
name = "base64ct"
version = "1.6.0"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "8c3c1a368f70d6cf7302d78f8f7093da241fb8e8807c05cc9e51a125895a6d5b"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
[[package]]
name = "bitflags"
2021-10-02 12:42:03 +02:00
version = "1.3.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
2021-10-02 12:42:03 +02:00
checksum = "bef38d45163c2f1dde094a7dfd33ccf595c92905c8f8f4fdc18d06fb1037718a"
[[package]]
name = "bitflags"
version = "2.0.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "487f1e0fcbe47deb8b0574e646def1c903389d95241dd1bbcc6ce4a715dfc0c1"
[[package]]
name = "blake3"
version = "1.3.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "42ae2468a89544a466886840aa467a25b766499f4f04bf7d9fcd10ecee9fccef"
dependencies = [
"arrayref",
"arrayvec",
"cc",
2021-10-02 12:42:03 +02:00
"cfg-if",
"constant_time_eq 0.2.5",
2022-11-30 03:32:27 +01:00
"digest",
]
2022-01-15 05:59:43 +01:00
[[package]]
name = "block-buffer"
version = "0.10.4"
2022-01-15 05:59:43 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "3078c7629b62d3f0439517fa394996acacc5cbc91c5a20d8c658e77abd503a71"
2022-01-15 05:59:43 +01:00
dependencies = [
"generic-array",
]
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
[[package]]
2022-11-30 03:32:27 +01:00
name = "block-padding"
version = "0.3.2"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
2022-11-30 03:32:27 +01:00
checksum = "0a90ec2df9600c28a01c56c4784c9207a96d2451833aeceb8cc97e4c9548bb78"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
dependencies = [
2022-11-30 03:32:27 +01:00
"generic-array",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
]
[[package]]
name = "bstr"
version = "0.2.17"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "ba3569f383e8f1598449f1a423e72e99569137b47740b1da11ef19af3d5c3223"
dependencies = [
"lazy_static",
"memchr",
"regex-automata",
"serde",
]
[[package]]
name = "bstr"
version = "1.4.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "c3d4260bcc2e8fc9df1eac4919a720effeb63a3f0952f5bf4944adfa18897f09"
dependencies = [
"memchr",
"once_cell",
"regex-automata",
"serde",
]
[[package]]
name = "bumpalo"
version = "3.12.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "0d261e256854913907f67ed06efbc3338dfe6179796deefc1ff763fc1aee5535"
[[package]]
name = "byteorder"
2021-03-27 05:47:16 +01:00
version = "1.4.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
2021-03-27 05:47:16 +01:00
checksum = "14c189c53d098945499cdfa7ecc63567cf3886b3332b312a5b4585d8d3a6a610"
[[package]]
name = "bytes"
version = "1.4.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "89b2fd2a0dcf38d7971e2194b6b6eebab45ae01067456a7fd93d5547a61b70be"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
[[package]]
name = "bzip2"
version = "0.4.4"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "bdb116a6ef3f6c3698828873ad02c3014b3c85cadb88496095628e3ef1e347f8"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
dependencies = [
"bzip2-sys",
"libc",
]
[[package]]
name = "bzip2-sys"
version = "0.1.11+1.0.8"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "736a955f3fa7875102d57c82b8cac37ec45224a07fd32d58f9f7a186b6cd4cdc"
dependencies = [
"cc",
"libc",
"pkg-config",
]
[[package]]
name = "camino"
version = "1.1.4"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "c530edf18f37068ac2d977409ed5cd50d53d73bc653c7647b48eb78976ac9ae2"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
[[package]]
name = "cast"
2022-09-24 04:39:21 +02:00
version = "0.3.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
2022-09-24 04:39:21 +02:00
checksum = "37b2a672a2cb129a2e41c10b1224bb368f9f37a2b16b612598138befd7b37eb5"
2022-11-30 03:32:27 +01:00
[[package]]
name = "cbc"
version = "0.1.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "26b52a9543ae338f279b96b0b9fed9c8093744685043739079ce85cd58f289a6"
dependencies = [
"cipher 0.4.4",
2022-11-30 03:32:27 +01:00
]
[[package]]
name = "cc"
version = "1.0.79"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "50d30906286121d95be3d479533b458f87493b30a4b5f79a607db8f5d11aa91f"
Backups (#1685) * Add zstd dep * Implement backend backup with zstd * Implement backup thinning * Write backup meta * Use new file ending anki21b * Asynchronously backup on collection close in Rust * Revert "Add zstd dep" This reverts commit 3fcb2141d2be15f907269d13275c41971431385c. * Add zstd again * Take backup col path from col struct * Fix formatting * Implement backup restoring on backend * Normalize restored media file names * Refactor `extract_legacy_data()` A bit cumbersome due to borrowing rules. * Refactor * Make thinning calendar-based and gradual * Consider last kept backups of previous stages * Import full apkgs and colpkgs with backend * Expose new backup settings * Test `BackupThinner` and make it deterministic * Mark backup_path when closing optional * Delete leaky timer * Add progress updates for restoring media * Write restored collection to tempfile first * Do collection compression in the background thread This has us currently storing an uncompressed and compressed copy of the collection in memory (not ideal), but means the collection can be closed without waiting for compression to complete. On a large collection, this takes a close and reopen from about 0.55s to about 0.07s. The old backup code for comparison: about 0.35s for compression off, about 8.5s for zip compression. * Use multithreading in zstd compression On my system, this reduces the compression time of a large collection from about 0.55s to 0.08s. * Stream compressed collection data into zip file * Tweak backup explanation + Fix incorrect tab order for ignore accents option * Decouple restoring backup and full import In the first case, no profile is opened, unless the new collection succeeds to load. In the second case, either the old collection is reloaded or the new one is loaded. * Fix number gap in Progress message * Don't revert backup when media fails but report it * Tweak error flow * Remove native BackupLimits enum * Fix type annotation * Add thinning test for whole year * Satisfy linter * Await async backup to finish * Move restart disclaimer out of backup tab Should be visible regardless of the current tab. * Write restored collection in chunks * Refactor * Write media in chunks and refactor * Log error if removing file fails * join_backup_task -> await_backup_completion * Refactor backup.rs * Refactor backup meta and collection extraction * Fix wrong error being returned * Call sync_all() on new collection * Add ImportError * Store logger in Backend, instead of creating one on demand init_backend() accepts a Logger rather than a log file, to allow other callers to customize the logger if they wish. In the future we may want to explore using the tracing crate as an alternative; it's a bit more ergonomic, as a logger doesn't need to be passed around, and it plays more nicely with async code. * Sync file contents prior to rename; sync folder after rename. * Limit backup creation to once per 30 min * Use zstd::stream::copy_decode * Make importing abortable * Don't revert if backup media is aborted * Set throttle implicitly * Change force flag to minimum_backup_interval * Don't attempt to open folders on Windows * Join last backup thread before starting new one Also refactor. * Disable auto sync and backup when restoring again * Force backup on full download * Include the reason why a media file import failed, and the file path - Introduce a FileIoError that contains a string representation of the underlying I/O error, and an associated path. There are a few places in the code where we're currently manually including the filename in a custom error message, and this is a step towards a more consistent approach (but we may be better served with a more general approach in the future similar to Anyhow's .context()) - Move the error message into importing.ftl, as it's a bit neater when error messages live in the same file as the rest of the messages associated with some functionality. * Fix importing of media files * Minor wording tweaks * Save an allocation I18n strings with replacements are already strings, so we can skip the extra allocation. Not that it matters here at all. * Terminate import if file missing from archive If a third-party tool is creating invalid archives, the user should know about it. This should be rare, so I did not attempt to make it translatable. * Skip multithreaded compression on small collections Co-authored-by: Damien Elmes <gpg@ankiweb.net>
2022-03-07 06:11:31 +01:00
dependencies = [
"jobserver",
]
[[package]]
name = "cfg-if"
version = "1.0.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "baf1de4339761588bc0619e3cbc0120ee582ebb74b53b4efbf79117bd2da40fd"
[[package]]
name = "chrono"
version = "0.4.24"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "4e3c5919066adf22df73762e50cffcde3a758f2a848b113b586d1f86728b673b"
dependencies = [
2022-09-24 04:39:21 +02:00
"iana-time-zone",
"num-integer",
"num-traits",
"winapi",
]
2022-09-24 04:39:21 +02:00
[[package]]
name = "ciborium"
version = "0.2.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "b0c137568cc60b904a7724001b35ce2630fd00d5d84805fbb608ab89509d788f"
dependencies = [
"ciborium-io",
"ciborium-ll",
"serde",
]
[[package]]
name = "ciborium-io"
version = "0.2.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "346de753af073cc87b52b2083a506b38ac176a44cfb05497b622e27be899b369"
[[package]]
name = "ciborium-ll"
version = "0.2.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "213030a2b5a4e0c0892b6652260cf6ccac84827b83a85a534e178e3906c4cf1b"
dependencies = [
"ciborium-io",
"half",
]
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
[[package]]
name = "cipher"
version = "0.3.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "7ee52072ec15386f770805afd189a01c8841be8696bed250fa2f13c4c0d6dfb7"
dependencies = [
"generic-array",
]
2022-11-30 03:32:27 +01:00
[[package]]
name = "cipher"
version = "0.4.4"
2022-11-30 03:32:27 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "773f3b9af64447d2ce9850330c473515014aa235e6a783b02db81ff39e4a3dad"
2022-11-30 03:32:27 +01:00
dependencies = [
"crypto-common",
"inout",
]
[[package]]
name = "clap"
version = "3.2.23"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "71655c45cb9845d3270c9d6df84ebe72b4dad3c2ba3f7023ad47c144e4e473a5"
dependencies = [
"bitflags 1.3.2",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
"clap_lex 0.2.4",
2022-09-24 04:39:21 +02:00
"indexmap",
"textwrap",
2022-09-24 04:39:21 +02:00
]
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
[[package]]
name = "clap"
version = "4.2.1"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "046ae530c528f252094e4a77886ee1374437744b2bff1497aa898bbddbbb29b3"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
dependencies = [
"clap_builder",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
"clap_derive",
"once_cell",
]
[[package]]
name = "clap_builder"
version = "4.2.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "223163f58c9a40c3b0a43e1c4b50a9ce09f007ea2cb1ec258a687945b4b7929f"
dependencies = [
"anstream",
"anstyle",
"bitflags 1.3.2",
"clap_lex 0.4.1",
"once_cell",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
"strsim",
"terminal_size",
]
[[package]]
name = "clap_complete"
version = "4.2.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "01c22dcfb410883764b29953103d9ef7bb8fe21b3fa1158bc99986c2067294bd"
dependencies = [
"clap 4.2.1",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
]
[[package]]
name = "clap_derive"
version = "4.2.0"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "3f9644cd56d6b87dbe899ef8b053e331c0637664e9e21a33dfcdc36093f5c5c4"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
dependencies = [
"heck",
"proc-macro2",
"quote",
"syn 2.0.12",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
]
2022-09-24 04:39:21 +02:00
[[package]]
name = "clap_lex"
version = "0.2.4"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "2850f2f5a82cbf437dd5af4d49848fbdfc27c157c3d010345776f952765261c5"
dependencies = [
"os_str_bytes",
]
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
[[package]]
name = "clap_lex"
version = "0.4.1"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "8a2dd5a6fe8c6e3502f568a6353e5273bbb15193ad9a89e457b9970798efbea1"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
[[package]]
name = "coarsetime"
version = "0.1.23"
2020-11-12 11:00:40 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "a90d114103adbc625300f346d4d09dfb4ab1c4a8df6868435dd903392ecf4354"
dependencies = [
"libc",
"once_cell",
"wasi 0.11.0+wasi-snapshot-preview1",
"wasm-bindgen",
]
[[package]]
name = "codespan"
version = "0.11.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "3362992a0d9f1dd7c3d0e89e0ab2bb540b7a95fea8cd798090e758fda2899b5e"
dependencies = [
"codespan-reporting",
"serde",
]
[[package]]
name = "codespan-reporting"
version = "0.11.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "3538270d33cc669650c4b093848450d380def10c331d38c768e34cac80576e6e"
dependencies = [
"serde",
"termcolor",
"unicode-width",
]
[[package]]
name = "concolor-override"
version = "1.0.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "a855d4a1978dc52fb0536a04d384c2c0c1aa273597f08b77c8c4d3b2eec6037f"
[[package]]
name = "concolor-query"
version = "0.3.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "88d11d52c3d7ca2e6d0040212be9e4dbbcd78b6447f535b6b561f449427944cf"
dependencies = [
"windows-sys 0.45.0",
]
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
[[package]]
name = "concurrent-queue"
version = "2.1.0"
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "c278839b831783b70278b14df4d45e1beb1aad306c07bb796637de9a0e323e8e"
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
dependencies = [
"crossbeam-utils",
]
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
[[package]]
name = "configure"
version = "0.0.0"
dependencies = [
"anyhow",
"itertools",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
"ninja_gen",
"workspace-hack",
]
[[package]]
name = "constant_time_eq"
version = "0.1.5"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "245097e9a4535ee1e3e3931fcfcd55a796a44c643e8596ff6566d68f09b87bbc"
[[package]]
name = "constant_time_eq"
version = "0.2.5"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "13418e745008f7349ec7e449155f419a61b92b58a99cc3616942b926825ec76b"
[[package]]
name = "convert_case"
version = "0.6.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "ec182b0ca2f35d8fc196cf3404988fd8b8c739a4d270ff118a398feb0cbec1ca"
dependencies = [
"unicode-segmentation",
]
2020-11-03 01:26:27 +01:00
[[package]]
name = "core-foundation"
version = "0.9.3"
2020-11-03 01:26:27 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "194a7a9e6de53fa55116934067c844d9d749312f75c6f6d0980e8c252f8c2146"
2020-11-03 01:26:27 +01:00
dependencies = [
"core-foundation-sys",
"libc",
]
[[package]]
name = "core-foundation-sys"
version = "0.8.3"
2020-11-03 01:26:27 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "5827cebf4670468b8772dd191856768aedcb1b0278a04f989f7766351917b9dc"
2020-11-03 01:26:27 +01:00
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
[[package]]
name = "cpufeatures"
version = "0.2.6"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "280a9f2d8b3a38871a3c8a46fb80db65e5e5ed97da80c4d08bf27fb63e35e181"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
dependencies = [
"libc",
]
[[package]]
name = "crc32fast"
version = "1.3.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "b540bd8bc810d3885c6ea91e2018302f68baba2129ab3e88f32389ee9370880d"
dependencies = [
2021-10-02 12:42:03 +02:00
"cfg-if",
]
[[package]]
name = "criterion"
2022-09-24 04:39:21 +02:00
version = "0.4.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
2022-09-24 04:39:21 +02:00
checksum = "e7c76e09c1aae2bc52b3d2f29e13c6572553b30c4aa1b8a49fd70de6412654cb"
dependencies = [
2022-09-24 04:39:21 +02:00
"anes",
"atty",
"cast",
2022-09-24 04:39:21 +02:00
"ciborium",
"clap 3.2.23",
"criterion-plot",
"itertools",
"lazy_static",
"num-traits",
"oorandom",
"plotters",
"rayon",
"regex",
"serde",
"serde_derive",
"serde_json",
"tinytemplate",
"walkdir",
]
[[package]]
name = "criterion-plot"
2022-09-24 04:39:21 +02:00
version = "0.5.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
2022-09-24 04:39:21 +02:00
checksum = "6b50826342786a51a89e2da3a28f1c32b06e387201bc2d19791f622c673706b1"
dependencies = [
"cast",
"itertools",
]
[[package]]
name = "crossbeam-channel"
2023-04-12 00:29:26 +02:00
version = "0.5.8"
source = "registry+https://github.com/rust-lang/crates.io-index"
2023-04-12 00:29:26 +02:00
checksum = "a33c2bf77f2df06183c3aa30d1e96c0695a313d4f9c453cc3762a6db39f99200"
dependencies = [
2021-10-02 12:42:03 +02:00
"cfg-if",
2021-02-03 11:29:48 +01:00
"crossbeam-utils",
]
[[package]]
name = "crossbeam-deque"
version = "0.8.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "ce6fd6f855243022dcecf8702fef0c297d4338e226845fe067f6341ad9fa0cef"
dependencies = [
"cfg-if",
"crossbeam-epoch",
"crossbeam-utils",
]
[[package]]
name = "crossbeam-epoch"
version = "0.9.14"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "46bd5f3f85273295a9d14aedfb86f6aadbff6d8f5295c4a9edb08e819dcf5695"
dependencies = [
2022-09-24 04:39:21 +02:00
"autocfg",
"cfg-if",
"crossbeam-utils",
"memoffset",
"scopeguard",
]
[[package]]
name = "crossbeam-utils"
version = "0.8.15"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "3c063cd8cc95f5c377ed0d4b49a4b21f632396ff690e8470c29b3359b346984b"
dependencies = [
2021-10-02 12:42:03 +02:00
"cfg-if",
]
2022-01-15 05:59:43 +01:00
[[package]]
name = "crypto-common"
2022-09-24 04:39:21 +02:00
version = "0.1.6"
2022-01-15 05:59:43 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
2022-09-24 04:39:21 +02:00
checksum = "1bfb12502f3fc46cca1bb51ac28df9d618d813cdc3d2f25b9fe775a34af26bb3"
2022-01-15 05:59:43 +01:00
dependencies = [
"generic-array",
"typenum",
2022-01-15 05:59:43 +01:00
]
CSV import/export fixes and features (#1898) * Fix footer moving upwards * Fix column detection Was broken because escaped line breaks were not considered. Also removes delimiter detection on `#columns:` line. User must use tabs or set delimiter beforehand. * Add CSV preview * Parse `#tags column:` * Optionally export deck and notetype with CSV * Avoid clones in CSV export * Prevent bottom of page appearing under footer (dae) * Increase padding to 1em (dae) With 0.5em, when a vertical scrollbar is shown, it sits right next to the right edge of the content, making it look like there's no right margin. * Experimental changes to make table fit+scroll (dae) - limit individual cells to 15em, and show ellipses when truncated - limit total table width to body width, so that inner table is shown with scrollbar - use class rather than id - ids are bad practice in Svelte components, as more than one may be displayed on a single page * Skip importing foreign notes with filtered decks Were implicitly imported into the default deck before. Also some refactoring to fetch deck ids and names beforehand. * Hide spacer below hidden field mapping * Fix guid being replaced when updating note * Fix dupe identity check Canonify tags before checking if dupe is identical, but only add update tags later if appropriate. * Fix deck export for notes with missing card 1 * Fix note lines starting with `#` csv crate doesn't support escaping a leading comment char. :( * Support import/export of guids * Strip HTML from preview rows * Fix initially set deck if current is filtered * Make isHtml toggle reactive * Fix `html_to_text_line()` stripping sound names * Tweak export option labels * Switch to patched rust-csv fork Fixes writing lines starting with `#`, so revert 5ece10ad05f331. * List column options with first column field * Fix flag for exports with HTML stripped
2022-06-09 02:28:01 +02:00
[[package]]
name = "csv"
version = "1.1.6"
source = "git+https://github.com/ankitects/rust-csv.git?rev=1c9d3aab6f79a7d815c69f925a46a4590c115f90#1c9d3aab6f79a7d815c69f925a46a4590c115f90"
dependencies = [
"bstr 0.2.17",
2022-09-24 04:39:21 +02:00
"csv-core",
"itoa",
CSV import/export fixes and features (#1898) * Fix footer moving upwards * Fix column detection Was broken because escaped line breaks were not considered. Also removes delimiter detection on `#columns:` line. User must use tabs or set delimiter beforehand. * Add CSV preview * Parse `#tags column:` * Optionally export deck and notetype with CSV * Avoid clones in CSV export * Prevent bottom of page appearing under footer (dae) * Increase padding to 1em (dae) With 0.5em, when a vertical scrollbar is shown, it sits right next to the right edge of the content, making it look like there's no right margin. * Experimental changes to make table fit+scroll (dae) - limit individual cells to 15em, and show ellipses when truncated - limit total table width to body width, so that inner table is shown with scrollbar - use class rather than id - ids are bad practice in Svelte components, as more than one may be displayed on a single page * Skip importing foreign notes with filtered decks Were implicitly imported into the default deck before. Also some refactoring to fetch deck ids and names beforehand. * Hide spacer below hidden field mapping * Fix guid being replaced when updating note * Fix dupe identity check Canonify tags before checking if dupe is identical, but only add update tags later if appropriate. * Fix deck export for notes with missing card 1 * Fix note lines starting with `#` csv crate doesn't support escaping a leading comment char. :( * Support import/export of guids * Strip HTML from preview rows * Fix initially set deck if current is filtered * Make isHtml toggle reactive * Fix `html_to_text_line()` stripping sound names * Tweak export option labels * Switch to patched rust-csv fork Fixes writing lines starting with `#`, so revert 5ece10ad05f331. * List column options with first column field * Fix flag for exports with HTML stripped
2022-06-09 02:28:01 +02:00
"ryu",
"serde",
]
[[package]]
name = "csv-core"
version = "0.1.10"
source = "git+https://github.com/ankitects/rust-csv.git?rev=1c9d3aab6f79a7d815c69f925a46a4590c115f90#1c9d3aab6f79a7d815c69f925a46a4590c115f90"
dependencies = [
"memchr",
]
[[package]]
name = "cxx"
version = "1.0.94"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "f61f1b6389c3fe1c316bf8a4dccc90a38208354b330925bce1f74a6c4756eb93"
dependencies = [
"cc",
"cxxbridge-flags",
"cxxbridge-macro",
"link-cplusplus",
]
[[package]]
name = "cxx-build"
version = "1.0.94"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "12cee708e8962df2aeb38f594aae5d827c022b6460ac71a7a3e2c3c2aae5a07b"
dependencies = [
"cc",
"codespan-reporting",
"once_cell",
"proc-macro2",
"quote",
"scratch",
"syn 2.0.12",
]
[[package]]
name = "cxxbridge-flags"
version = "1.0.94"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "7944172ae7e4068c533afbb984114a56c46e9ccddda550499caa222902c7f7bb"
[[package]]
name = "cxxbridge-macro"
version = "1.0.94"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "2345488264226bf682893e25de0769f3360aac9957980ec49361b083ddaa5bc5"
dependencies = [
"proc-macro2",
"quote",
"syn 2.0.12",
]
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
[[package]]
name = "deadpool"
version = "0.9.5"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "421fe0f90f2ab22016f32a9881be5134fdd71c65298917084b0c7477cbc3856e"
dependencies = [
"async-trait",
"deadpool-runtime",
"num_cpus",
"retain_mut",
"tokio",
]
[[package]]
name = "deadpool-runtime"
version = "0.1.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "eaa37046cc0f6c3cc6090fbdbf73ef0b8ef4cfcc37f6befc0020f63e8cf121e1"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
[[package]]
name = "des"
2022-11-30 03:32:27 +01:00
version = "0.8.1"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
2022-11-30 03:32:27 +01:00
checksum = "ffdd80ce8ce993de27e9f063a444a4d53ce8e8db4c1f00cc03af5ad5a9867a1e"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
dependencies = [
"cipher 0.4.4",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
]
[[package]]
name = "difflib"
version = "0.4.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "6184e33543162437515c2e2b48714794e37845ec9851711914eec9d308f6ebe8"
[[package]]
name = "digest"
version = "0.10.6"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "8168378f4e5023e7218c89c891c0fd8ecdb5e5e4f18cb78f38cf245dd021e76f"
dependencies = [
2022-11-30 03:32:27 +01:00
"block-buffer",
2022-01-15 05:59:43 +01:00
"crypto-common",
"subtle",
]
[[package]]
name = "displaydoc"
version = "0.2.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "3bf95dc3f046b9da4f2d51833c0d3547d8564ef6910f5c1ed130306a75b92886"
dependencies = [
"proc-macro2",
"quote",
"syn 1.0.109",
]
[[package]]
name = "doc-comment"
version = "0.3.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "fea41bba32d969b513997752735605054bc0dfa92b4c56bf1189f2e174be7a10"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
[[package]]
name = "duct"
version = "0.13.6"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "37ae3fc31835f74c2a7ceda3aeede378b0ae2e74c8f1c36559fcc9ae2a4e7d3e"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
dependencies = [
"libc",
"once_cell",
"os_pipe",
"shared_child",
]
[[package]]
name = "dunce"
version = "1.0.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "0bd4b30a6560bbd9b4620f4de34c3f14f60848e58a9b7216801afcb4c7b31c3c"
[[package]]
name = "either"
version = "1.8.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "7fcaabb2fef8c910e7f4c7ce9f67a1283a1715879a7c230ca9d6d1ae31f16d91"
[[package]]
name = "elasticlunr-rs"
version = "3.0.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "41e83863a500656dfa214fee6682de9c5b9f03de6860fec531235ed2ae9f6571"
dependencies = [
"regex",
"serde",
"serde_derive",
"serde_json",
]
[[package]]
name = "encoding_rs"
version = "0.8.32"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "071a31f4ee85403370b58aca746f01041ede6f0da2730960ad001edc2b71b394"
dependencies = [
2021-10-02 12:42:03 +02:00
"cfg-if",
]
[[package]]
name = "env_logger"
2022-11-30 03:32:27 +01:00
version = "0.10.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
2022-11-30 03:32:27 +01:00
checksum = "85cdab6a89accf66733ad5a1693a4dcced6aeff64602b634530dd73c1f3ee9f0"
dependencies = [
"humantime",
2022-11-30 03:32:27 +01:00
"is-terminal",
"log",
"regex",
"termcolor",
]
[[package]]
name = "errno"
version = "0.3.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "4bcfec3a70f97c962c307b2d2c56e358cf1d00b558d74262b5f929ee8cc7e73a"
dependencies = [
"errno-dragonfly",
"libc",
"windows-sys 0.48.0",
]
[[package]]
name = "errno-dragonfly"
version = "0.1.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "aa68f1b12764fab894d2755d2518754e71b4fd80ecfb822714a1206c2aab39bf"
dependencies = [
"cc",
"libc",
]
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
[[package]]
name = "event-listener"
version = "2.5.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "0206175f82b8d6bf6652ff7d71a1e27fd2e4efde587fd368662814d6ec1d9ce0"
[[package]]
name = "fallible-iterator"
version = "0.2.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "4443176a9f2c162692bd3d352d745ef9413eec5782a80d8fd6f8a1ac692a07f7"
[[package]]
name = "fallible-streaming-iterator"
version = "0.1.9"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "7360491ce676a36bf9bb3c56c1aa791658183a54d2744120f27285738d90465a"
2022-01-15 05:59:43 +01:00
[[package]]
name = "fastrand"
version = "1.9.0"
2022-01-15 05:59:43 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "e51093e27b0797c359783294ca4f0a911c270184cb10f85783b118614a1501be"
2022-01-15 05:59:43 +01:00
dependencies = [
"instant",
]
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
[[package]]
name = "filetime"
version = "0.2.20"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "8a3de6e8d11b22ff9edc6d916f890800597d60f8b2da1caf2955c274638d6412"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
dependencies = [
"cfg-if",
"libc",
"redox_syscall 0.2.16",
"windows-sys 0.45.0",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
]
[[package]]
name = "find-winsdk"
version = "0.2.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "a8cbf17b871570c1f8612b763bac3e86290602bcf5dc3c5ce657e0e1e9071d9e"
dependencies = [
"serde",
"serde_derive",
"winreg 0.5.1",
]
[[package]]
name = "fixedbitset"
2022-09-24 04:39:21 +02:00
version = "0.4.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
2022-09-24 04:39:21 +02:00
checksum = "0ce7134b9999ecaf8bcd65542e436736ef32ddca1b3e06094cb6ec5755203b80"
[[package]]
name = "flate2"
version = "1.0.25"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "a8a2db397cb1c8772f31494cb8917e48cd1e64f0fa7efac59fbd741a0a8ce841"
dependencies = [
"crc32fast",
"miniz_oxide",
]
[[package]]
name = "fluent"
2021-10-02 12:42:03 +02:00
version = "0.16.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
2021-10-02 12:42:03 +02:00
checksum = "61f69378194459db76abd2ce3952b790db103ceb003008d3d50d97c41ff847a7"
dependencies = [
"fluent-bundle",
"unic-langid",
]
[[package]]
name = "fluent-bundle"
version = "0.15.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "e242c601dec9711505f6d5bbff5bedd4b61b2469f2e8bb8e57ee7c9747a87ffd"
dependencies = [
"fluent-langneg",
"fluent-syntax",
"intl-memoizer",
"intl_pluralrules",
2021-03-27 04:24:11 +01:00
"rustc-hash",
"self_cell",
"smallvec",
"unic-langid",
]
[[package]]
name = "fluent-langneg"
version = "0.13.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "2c4ad0989667548f06ccd0e306ed56b61bd4d35458d54df5ec7587c0e8ed5e94"
dependencies = [
"unic-langid",
]
[[package]]
name = "fluent-syntax"
2021-03-27 04:24:11 +01:00
version = "0.11.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
2021-03-27 04:24:11 +01:00
checksum = "c0abed97648395c902868fee9026de96483933faa54ea3b40d652f7dfe61ca78"
dependencies = [
"thiserror",
]
[[package]]
name = "fnv"
version = "1.0.7"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "3f9eec918d3f24069decb9af1554cad7c880e2da24a9afd88aca000531ab82c1"
[[package]]
name = "foreign-types"
version = "0.3.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "f6f339eb8adc052cd2ca78910fda869aefa38d22d5cb648e6485e4d3fc06f3b1"
dependencies = [
"foreign-types-shared",
]
[[package]]
name = "foreign-types-shared"
version = "0.1.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "00b0228411908ca8685dba7fc2cdd70ec9990a6e753e89b6ac91a84c40fbaf4b"
[[package]]
name = "form_urlencoded"
2022-09-24 04:39:21 +02:00
version = "1.1.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
2022-09-24 04:39:21 +02:00
checksum = "a9c384f161156f5260c24a097c56119f9be8c798586aecc13afbcbe7b7e26bf8"
dependencies = [
"percent-encoding",
]
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
[[package]]
name = "forwarded-header-value"
version = "0.1.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "8835f84f38484cc86f110a805655697908257fb9a7af005234060891557198e9"
dependencies = [
"nonempty",
"thiserror",
]
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
[[package]]
name = "fs2"
version = "0.4.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "9564fc758e15025b46aa6643b1b77d047d1a56a1aea6e01002ac0c7026876213"
dependencies = [
"libc",
"winapi",
]
[[package]]
name = "fsevent-sys"
version = "4.1.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "76ee7a02da4d231650c7cea31349b889be2f45ddb3ef3032d2ec8185f6313fd2"
dependencies = [
"libc",
]
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
[[package]]
name = "ftl"
version = "0.0.0"
dependencies = [
"camino",
"snafu",
"workspace-hack",
]
[[package]]
name = "futf"
version = "0.1.5"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "df420e2e84819663797d1ec6544b13c5be84629e7bb00dc960d6917db2987843"
dependencies = [
"mac",
"new_debug_unreachable",
]
[[package]]
name = "futures"
version = "0.3.28"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "23342abe12aba583913b2e62f22225ff9c950774065e4bfb61a19cd9770fec40"
dependencies = [
"futures-channel",
"futures-core",
"futures-executor",
"futures-io",
"futures-sink",
"futures-task",
"futures-util",
]
[[package]]
name = "futures-channel"
version = "0.3.28"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "955518d47e09b25bbebc7a18df10b81f0c766eaf4c4f1cccef2fca5f2a4fb5f2"
dependencies = [
"futures-core",
"futures-sink",
]
[[package]]
name = "futures-core"
version = "0.3.28"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "4bca583b7e26f571124fe5b7561d49cb2868d79116cfa0eefce955557c6fee8c"
[[package]]
name = "futures-executor"
version = "0.3.28"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "ccecee823288125bd88b4d7f565c9e58e41858e47ab72e8ea2d64e93624386e0"
dependencies = [
"futures-core",
"futures-task",
"futures-util",
]
[[package]]
name = "futures-io"
version = "0.3.28"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "4fff74096e71ed47f8e023204cfd0aa1289cd54ae5430a9523be060cdb849964"
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
[[package]]
name = "futures-lite"
version = "1.12.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "7694489acd39452c77daa48516b894c153f192c3578d5a839b62c58099fcbf48"
dependencies = [
"fastrand",
"futures-core",
"futures-io",
"memchr",
"parking",
"pin-project-lite",
"waker-fn",
]
[[package]]
name = "futures-macro"
version = "0.3.28"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "89ca545a94061b6365f2c7355b4b32bd20df3ff95f02da9329b34ccc3bd6ee72"
dependencies = [
"proc-macro2",
"quote",
"syn 2.0.12",
]
[[package]]
name = "futures-sink"
version = "0.3.28"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "f43be4fe21a13b9781a69afa4985b0f6ee0e1afab2c6f454a8cf30e2b2237b6e"
[[package]]
name = "futures-task"
version = "0.3.28"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "76d3d132be6c0e6aa1534069c705a74a5997a356c0dc2f86a47765e5617c5b65"
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
[[package]]
name = "futures-timer"
version = "3.0.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "e64b03909df88034c26dc1547e8970b91f98bdb65165d6a4e9110d94263dbb2c"
[[package]]
name = "futures-util"
version = "0.3.28"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "26b01e40b772d54cf6c6d721c1d1abd0647a0106a12ecaa1c186273392a69533"
dependencies = [
"futures-channel",
"futures-core",
"futures-io",
"futures-macro",
"futures-sink",
"futures-task",
"memchr",
"pin-project-lite",
"pin-utils",
"slab",
]
[[package]]
name = "generic-array"
version = "0.14.7"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "85649ca51fd72272d7821adaf274ad91c288277713d9c18820d8499a7ff69e9a"
dependencies = [
"typenum",
"version_check",
]
[[package]]
name = "getopts"
version = "0.2.21"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "14dbbfd5c71d70241ecf9e6f13737f7b5ce823821063188d7e46c41d371eebd5"
dependencies = [
"unicode-width",
]
[[package]]
name = "getrandom"
2021-01-05 10:58:53 +01:00
version = "0.1.16"
source = "registry+https://github.com/rust-lang/crates.io-index"
2021-01-05 10:58:53 +01:00
checksum = "8fc3cb4d91f53b50155bdcfd23f6a4c39ae1969c2ae85982b135750cccaf5fce"
dependencies = [
2021-10-02 12:42:03 +02:00
"cfg-if",
"libc",
"wasi 0.9.0+wasi-snapshot-preview1",
]
2021-02-03 11:29:48 +01:00
[[package]]
name = "getrandom"
version = "0.2.8"
2021-02-03 11:29:48 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "c05aeb6a22b8f62540c194aac980f2115af067bfe15a0734d7277a768d396b31"
2021-02-03 11:29:48 +01:00
dependencies = [
2021-10-02 12:42:03 +02:00
"cfg-if",
2021-02-03 11:29:48 +01:00
"libc",
2022-09-24 04:39:21 +02:00
"wasi 0.11.0+wasi-snapshot-preview1",
2021-02-03 11:29:48 +01:00
]
[[package]]
name = "gimli"
version = "0.27.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "ad0a93d233ebf96623465aad4046a8d3aa4da22d4f4beba5388838c8a434bbb4"
[[package]]
name = "gitignore"
version = "1.0.7"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "78aa90e4620c1498ac434c06ba6e521b525794bbdacf085d490cc794b4a2f9a4"
dependencies = [
"glob",
]
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
[[package]]
name = "glob"
version = "0.3.1"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "d2fabcfbdc87f4758337ca535fb41a6d701b65693ce38287d856d1674551ec9b"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
[[package]]
name = "globset"
version = "0.4.10"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "029d74589adefde59de1a0c4f4732695c32805624aec7b68d91503d4dba79afc"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
dependencies = [
"aho-corasick",
"bstr 1.4.0",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
"fnv",
"log",
"regex",
]
[[package]]
name = "h2"
version = "0.3.18"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "17f8a914c2987b688368b5138aa05321db91f4090cf26118185672ad588bce21"
dependencies = [
"bytes",
"fnv",
"futures-core",
"futures-sink",
"futures-util",
"http",
"indexmap",
"slab",
"tokio",
"tokio-util",
"tracing",
]
[[package]]
name = "half"
version = "1.8.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "eabb4a44450da02c90444cf74558da904edde8fb4e9035a9a6a4e15445af0bd7"
[[package]]
name = "handlebars"
version = "4.3.6"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "035ef95d03713f2c347a72547b7cd38cbc9af7cd51e6099fb62d586d4a6dee3a"
dependencies = [
"log",
"pest",
"pest_derive",
"serde",
"serde_json",
"thiserror",
]
2021-06-25 08:22:21 +02:00
[[package]]
name = "hashbrown"
2022-09-24 04:39:21 +02:00
version = "0.12.3"
2021-06-25 08:22:21 +02:00
source = "registry+https://github.com/rust-lang/crates.io-index"
2022-09-24 04:39:21 +02:00
checksum = "8a9ee70c43aaf417c914396645a0fa852624801b24ebb7ae78fe8272889ac888"
dependencies = [
"ahash",
]
[[package]]
name = "hashlink"
2022-09-24 04:39:21 +02:00
version = "0.8.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
2022-09-24 04:39:21 +02:00
checksum = "69fe1fcf8b4278d860ad0548329f892a3631fb63f82574df68275f34cdbe0ffa"
dependencies = [
2021-10-02 12:42:03 +02:00
"hashbrown",
]
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
[[package]]
name = "headers"
version = "0.3.8"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "f3e372db8e5c0d213e0cd0b9be18be2aca3d44cf2fe30a9d46a65581cd454584"
dependencies = [
"base64 0.13.1",
"bitflags 1.3.2",
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
"bytes",
"headers-core",
"http",
"httpdate",
"mime",
"sha1",
]
[[package]]
name = "headers-core"
version = "0.2.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "e7f66481bfee273957b1f20485a4ff3362987f85b2c236580d81b4eb7a326429"
dependencies = [
"http",
]
[[package]]
name = "heck"
version = "0.4.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "95505c38b4572b2d910cecb0281560f54b440a19336cbbcb27bf6ce6adc6f5a8"
[[package]]
name = "hermit-abi"
2021-06-25 07:35:25 +02:00
version = "0.1.19"
source = "registry+https://github.com/rust-lang/crates.io-index"
2021-06-25 07:35:25 +02:00
checksum = "62b467343b94ba476dcb2500d242dadbb39557df889310ac77c5d99100aaac33"
dependencies = [
"libc",
]
[[package]]
name = "hermit-abi"
version = "0.2.6"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "ee512640fe35acbfb4bb779db6f0d80704c2cacfa2e39b601ef3e3f47d1ae4c7"
dependencies = [
"libc",
]
[[package]]
name = "hermit-abi"
version = "0.3.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "fed44880c466736ef9a5c5b5facefb5ed0785676d0c02d612db14e54f0d84286"
[[package]]
name = "hex"
2021-03-07 10:04:34 +01:00
version = "0.4.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
2021-03-07 10:04:34 +01:00
checksum = "7f24254aa9a54b5c858eaee2f5bccdb46aaf0e486a595ed5fd8f86ba55232a70"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
[[package]]
name = "hmac"
version = "0.12.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "6c49c37c09c17a53d937dfbb742eb3a961d65a994e6bcdcf37e7399d0cc8ab5e"
dependencies = [
2022-11-30 03:32:27 +01:00
"digest",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
]
2022-09-24 04:39:21 +02:00
[[package]]
name = "html5ever"
version = "0.26.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
2022-09-24 04:39:21 +02:00
checksum = "bea68cab48b8459f17cf1c944c67ddc572d272d9f2b274140f223ecb1da4a3b7"
dependencies = [
"log",
"mac",
"markup5ever",
"proc-macro2",
"quote",
"syn 1.0.109",
]
[[package]]
name = "htmlescape"
version = "0.3.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "e9025058dae765dee5070ec375f591e2ba14638c63feff74f13805a72e523163"
[[package]]
name = "http"
version = "0.2.9"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "bd6effc99afb63425aff9b05836f029929e345a6148a14b7ecd5ab67af944482"
dependencies = [
"bytes",
"fnv",
"itoa",
]
[[package]]
name = "http-body"
2022-09-24 04:39:21 +02:00
version = "0.4.5"
source = "registry+https://github.com/rust-lang/crates.io-index"
2022-09-24 04:39:21 +02:00
checksum = "d5f38f16d184e36f2408a55281cd658ecbd3ca05cce6d6510a176eca393e26d1"
dependencies = [
"bytes",
"http",
"pin-project-lite",
]
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
[[package]]
name = "http-range-header"
version = "0.3.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "0bfe8eed0a9285ef776bb792479ea3834e8b94e13d615c2f66d03dd50a435a29"
[[package]]
name = "http-types"
version = "2.12.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "6e9b187a72d63adbfba487f48095306ac823049cb504ee195541e91c7775f5ad"
dependencies = [
"anyhow",
"async-channel",
"base64 0.13.1",
"futures-lite",
"http",
"infer",
"pin-project-lite",
"rand 0.7.3",
"serde",
"serde_json",
"serde_qs",
"serde_urlencoded",
"url",
]
[[package]]
name = "httparse"
2022-09-24 04:39:21 +02:00
version = "1.8.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
2022-09-24 04:39:21 +02:00
checksum = "d897f394bad6a705d5f4104762e116a75639e470d80901eed05a860a95cb1904"
[[package]]
name = "httpdate"
2021-11-18 11:54:00 +01:00
version = "1.0.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
2021-11-18 11:54:00 +01:00
checksum = "c4a1e36c821dbe04574f602848a19f742f4fb3c98d40449f11bcad18d6b17421"
[[package]]
name = "humantime"
2021-02-03 11:29:48 +01:00
version = "2.1.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
2021-02-03 11:29:48 +01:00
checksum = "9a3a5bfb195931eeb336b2a7b4d761daec841b97f947d34394601737a7bba5e4"
[[package]]
name = "hyper"
version = "0.14.25"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "cc5e554ff619822309ffd57d8734d77cd5ce6238bc956f037ea06c58238c9899"
dependencies = [
"bytes",
"futures-channel",
"futures-core",
"futures-util",
"h2",
"http",
"http-body",
"httparse",
"httpdate",
"itoa",
2021-06-16 08:10:57 +02:00
"pin-project-lite",
"socket2",
"tokio",
"tower-service",
"tracing",
"want",
]
[[package]]
name = "hyper-rustls"
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
version = "0.23.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
checksum = "1788965e61b367cd03a62950836d5cd41560c3577d90e40e0819373194d1661c"
dependencies = [
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
"http",
"hyper",
"rustls",
"tokio",
"tokio-rustls",
]
[[package]]
name = "hyper-tls"
version = "0.5.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "d6183ddfa99b85da61a140bea0efc93fdf56ceaa041b37d553518030827f9905"
dependencies = [
"bytes",
"hyper",
"native-tls",
"tokio",
"tokio-native-tls",
]
2022-09-24 04:39:21 +02:00
[[package]]
name = "iana-time-zone"
version = "0.1.56"
2022-09-24 04:39:21 +02:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "0722cd7114b7de04316e7ea5456a0bbb20e4adb46fd27a3697adb812cff0f37c"
2022-09-24 04:39:21 +02:00
dependencies = [
"android_system_properties",
"core-foundation-sys",
"iana-time-zone-haiku",
2022-09-24 04:39:21 +02:00
"js-sys",
"wasm-bindgen",
"windows",
2022-09-24 04:39:21 +02:00
]
[[package]]
name = "iana-time-zone-haiku"
version = "0.1.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "0703ae284fc167426161c2e3f1da3ea71d94b21bedbcc9494e92b28e334e3dca"
dependencies = [
"cxx",
"cxx-build",
]
V3 parent limits (#1638) * avoid repinning Rust deps by default * add id_tree dependency * Respect intermediate child limits in v3 * Test new behaviour of v3 counts * Rework v3 queue building to respect parent limits * Add missing did field to SQL query * Fix `LimitTreeMap::is_exhausted()` * Rework tree building logic https://github.com/ankitects/anki/pull/1638#discussion_r798328734 * Add timer for build_queues() * `is_exhausted()` -> `limit_reached()` * Move context and limits into `QueueBuilder` This allows for moving more logic into QueueBuilder, so less passing around of arguments. Unfortunately, some tests will require additional work to set up. * Fix stop condition in new_cards_by_position * Fix order gather order of new cards by deck * Add scheduler/queue/builder/burying.rs * Fix bad tree due to unsorted child decks * Fix comment * Fix `cap_new_to_review_rec()` * Add test for new card gathering * Always sort `child_decks()` * Fix deck removal in `cap_new_to_review_rec()` * Fix sibling ordering in new card gathering * Remove limits for deck total count with children * Add random gather order * Remove bad sibling order handling All routines ensure ascending order now. Also do some other minor refactoring. * Remove queue truncating All routines stop now as soon as the root limit is reached. * Move deck fetching into `QueueBuilder::new()` * Rework new card gather and sort options https://github.com/ankitects/anki/pull/1638#issuecomment-1032173013 * Disable new sort order choices ... depending on set gather order. * Use enum instead of numbers * Ensure valid sort order setting * Update new gather and sort order tooltips * Warn about random insertion order with v3 * Revert "Add timer for build_queues()" This reverts commit c9f5fc6ebe87953c17a0c842990b009b5596c69c. * Update rslib/src/storage/card/mod.rs (dae) * minor wording tweaks to the tooltips (dae) + move legacy strings to bottom + consistent capitalization (our leech action still needs fixing, but that will require introducing a new 'suspend card' string as the existing one is used elsewhere as well)
2022-02-10 00:55:43 +01:00
[[package]]
name = "id_tree"
version = "1.8.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "bcd9db8dd5be8bde5a2624ed4b2dfb74368fe7999eb9c4940fd3ca344b61071a"
dependencies = [
"snowflake",
]
[[package]]
name = "idna"
2022-09-24 04:39:21 +02:00
version = "0.3.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
2022-09-24 04:39:21 +02:00
checksum = "e14ddfc70884202db2244c223200c204c2bda1bc6e0998d11b5e024d657209e6"
dependencies = [
"unicode-bidi",
"unicode-normalization",
]
[[package]]
name = "indexmap"
version = "1.9.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "bd070e393353796e801d209ad339e89596eb4c8d430d18ede6a1cced8fafbd99"
dependencies = [
"autocfg",
2021-10-02 12:42:03 +02:00
"hashbrown",
]
[[package]]
name = "indoc"
version = "1.0.9"
2021-03-07 10:04:34 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "bfa799dd5ed20a7e349f3b4639aa80d74549c81716d9ec4f994c9b5815598306"
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
[[package]]
name = "infer"
version = "0.2.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "64e9829a50b42bb782c1df523f78d332fe371b10c661e78b7a3c34b0198e9fac"
[[package]]
name = "inflections"
version = "1.1.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "a257582fdcde896fd96463bf2d40eefea0580021c0712a0e2b028b60b47a837a"
[[package]]
name = "inotify"
version = "0.9.6"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "f8069d3ec154eb856955c1c0fbffefbf5f3c40a104ec912d4797314c1801abff"
dependencies = [
"bitflags 1.3.2",
"inotify-sys",
"libc",
]
[[package]]
name = "inotify-sys"
version = "0.1.5"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "e05c02b5e89bff3b946cedeca278abc628fe811e604f027c45a8aa3cf793d0eb"
dependencies = [
"libc",
]
2022-11-30 03:32:27 +01:00
[[package]]
name = "inout"
version = "0.1.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "a0c10553d664a4d0bcff9f4215d0aac67a639cc68ef660840afe309b807bc9f5"
dependencies = [
"block-padding",
"generic-array",
]
[[package]]
name = "instant"
version = "0.1.12"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "7a5bbe824c507c5da5956355e86a746d82e0e1464f65d862cc5e71da70e94b2c"
dependencies = [
2021-10-02 12:42:03 +02:00
"cfg-if",
]
[[package]]
name = "intl-memoizer"
2021-02-03 11:29:48 +01:00
version = "0.5.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
2021-02-03 11:29:48 +01:00
checksum = "c310433e4a310918d6ed9243542a6b83ec1183df95dff8f23f87bb88a264a66f"
dependencies = [
"type-map",
"unic-langid",
]
[[package]]
name = "intl_pluralrules"
version = "7.0.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "078ea7b7c29a2b4df841a7f6ac8775ff6074020c6776d48491ce2268e068f972"
dependencies = [
"unic-langid",
]
[[package]]
name = "io-lifetimes"
version = "1.0.11"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "eae7b9aee968036d54dce06cebaefd919e4472e753296daccd6d344e3e2df0c2"
dependencies = [
"hermit-abi 0.3.1",
"libc",
"windows-sys 0.48.0",
]
2020-11-03 01:26:27 +01:00
[[package]]
name = "ipnet"
version = "2.7.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "12b6ee2129af8d4fb011108c73d99a1b83a85977f23b82460c0ae2e25bb4b57f"
[[package]]
name = "is-terminal"
version = "0.4.6"
2020-11-03 01:26:27 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "256017f749ab3117e93acb91063009e1f1bb56d03965b14c2c8df4eb02c524d8"
dependencies = [
"hermit-abi 0.3.1",
"io-lifetimes",
"rustix",
"windows-sys 0.45.0",
]
2020-11-03 01:26:27 +01:00
2021-06-25 07:35:25 +02:00
[[package]]
name = "itertools"
2022-09-24 04:39:21 +02:00
version = "0.10.5"
2021-06-25 07:35:25 +02:00
source = "registry+https://github.com/rust-lang/crates.io-index"
2022-09-24 04:39:21 +02:00
checksum = "b0fd2260e829bddf4cb6ea802289de2f86d6a7a690192fbe91b3f46e0f2c8473"
2021-06-25 07:35:25 +02:00
dependencies = [
"either",
]
2022-01-15 05:59:43 +01:00
[[package]]
name = "itoa"
version = "1.0.6"
2022-01-15 05:59:43 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "453ad9f582a441959e5f0d088b02ce04cfe8d51a8eaf077f12ac6d3e94164ca6"
2022-01-15 05:59:43 +01:00
Backups (#1685) * Add zstd dep * Implement backend backup with zstd * Implement backup thinning * Write backup meta * Use new file ending anki21b * Asynchronously backup on collection close in Rust * Revert "Add zstd dep" This reverts commit 3fcb2141d2be15f907269d13275c41971431385c. * Add zstd again * Take backup col path from col struct * Fix formatting * Implement backup restoring on backend * Normalize restored media file names * Refactor `extract_legacy_data()` A bit cumbersome due to borrowing rules. * Refactor * Make thinning calendar-based and gradual * Consider last kept backups of previous stages * Import full apkgs and colpkgs with backend * Expose new backup settings * Test `BackupThinner` and make it deterministic * Mark backup_path when closing optional * Delete leaky timer * Add progress updates for restoring media * Write restored collection to tempfile first * Do collection compression in the background thread This has us currently storing an uncompressed and compressed copy of the collection in memory (not ideal), but means the collection can be closed without waiting for compression to complete. On a large collection, this takes a close and reopen from about 0.55s to about 0.07s. The old backup code for comparison: about 0.35s for compression off, about 8.5s for zip compression. * Use multithreading in zstd compression On my system, this reduces the compression time of a large collection from about 0.55s to 0.08s. * Stream compressed collection data into zip file * Tweak backup explanation + Fix incorrect tab order for ignore accents option * Decouple restoring backup and full import In the first case, no profile is opened, unless the new collection succeeds to load. In the second case, either the old collection is reloaded or the new one is loaded. * Fix number gap in Progress message * Don't revert backup when media fails but report it * Tweak error flow * Remove native BackupLimits enum * Fix type annotation * Add thinning test for whole year * Satisfy linter * Await async backup to finish * Move restart disclaimer out of backup tab Should be visible regardless of the current tab. * Write restored collection in chunks * Refactor * Write media in chunks and refactor * Log error if removing file fails * join_backup_task -> await_backup_completion * Refactor backup.rs * Refactor backup meta and collection extraction * Fix wrong error being returned * Call sync_all() on new collection * Add ImportError * Store logger in Backend, instead of creating one on demand init_backend() accepts a Logger rather than a log file, to allow other callers to customize the logger if they wish. In the future we may want to explore using the tracing crate as an alternative; it's a bit more ergonomic, as a logger doesn't need to be passed around, and it plays more nicely with async code. * Sync file contents prior to rename; sync folder after rename. * Limit backup creation to once per 30 min * Use zstd::stream::copy_decode * Make importing abortable * Don't revert if backup media is aborted * Set throttle implicitly * Change force flag to minimum_backup_interval * Don't attempt to open folders on Windows * Join last backup thread before starting new one Also refactor. * Disable auto sync and backup when restoring again * Force backup on full download * Include the reason why a media file import failed, and the file path - Introduce a FileIoError that contains a string representation of the underlying I/O error, and an associated path. There are a few places in the code where we're currently manually including the filename in a custom error message, and this is a step towards a more consistent approach (but we may be better served with a more general approach in the future similar to Anyhow's .context()) - Move the error message into importing.ftl, as it's a bit neater when error messages live in the same file as the rest of the messages associated with some functionality. * Fix importing of media files * Minor wording tweaks * Save an allocation I18n strings with replacements are already strings, so we can skip the extra allocation. Not that it matters here at all. * Terminate import if file missing from archive If a third-party tool is creating invalid archives, the user should know about it. This should be rare, so I did not attempt to make it translatable. * Skip multithreaded compression on small collections Co-authored-by: Damien Elmes <gpg@ankiweb.net>
2022-03-07 06:11:31 +01:00
[[package]]
name = "jobserver"
version = "0.1.26"
Backups (#1685) * Add zstd dep * Implement backend backup with zstd * Implement backup thinning * Write backup meta * Use new file ending anki21b * Asynchronously backup on collection close in Rust * Revert "Add zstd dep" This reverts commit 3fcb2141d2be15f907269d13275c41971431385c. * Add zstd again * Take backup col path from col struct * Fix formatting * Implement backup restoring on backend * Normalize restored media file names * Refactor `extract_legacy_data()` A bit cumbersome due to borrowing rules. * Refactor * Make thinning calendar-based and gradual * Consider last kept backups of previous stages * Import full apkgs and colpkgs with backend * Expose new backup settings * Test `BackupThinner` and make it deterministic * Mark backup_path when closing optional * Delete leaky timer * Add progress updates for restoring media * Write restored collection to tempfile first * Do collection compression in the background thread This has us currently storing an uncompressed and compressed copy of the collection in memory (not ideal), but means the collection can be closed without waiting for compression to complete. On a large collection, this takes a close and reopen from about 0.55s to about 0.07s. The old backup code for comparison: about 0.35s for compression off, about 8.5s for zip compression. * Use multithreading in zstd compression On my system, this reduces the compression time of a large collection from about 0.55s to 0.08s. * Stream compressed collection data into zip file * Tweak backup explanation + Fix incorrect tab order for ignore accents option * Decouple restoring backup and full import In the first case, no profile is opened, unless the new collection succeeds to load. In the second case, either the old collection is reloaded or the new one is loaded. * Fix number gap in Progress message * Don't revert backup when media fails but report it * Tweak error flow * Remove native BackupLimits enum * Fix type annotation * Add thinning test for whole year * Satisfy linter * Await async backup to finish * Move restart disclaimer out of backup tab Should be visible regardless of the current tab. * Write restored collection in chunks * Refactor * Write media in chunks and refactor * Log error if removing file fails * join_backup_task -> await_backup_completion * Refactor backup.rs * Refactor backup meta and collection extraction * Fix wrong error being returned * Call sync_all() on new collection * Add ImportError * Store logger in Backend, instead of creating one on demand init_backend() accepts a Logger rather than a log file, to allow other callers to customize the logger if they wish. In the future we may want to explore using the tracing crate as an alternative; it's a bit more ergonomic, as a logger doesn't need to be passed around, and it plays more nicely with async code. * Sync file contents prior to rename; sync folder after rename. * Limit backup creation to once per 30 min * Use zstd::stream::copy_decode * Make importing abortable * Don't revert if backup media is aborted * Set throttle implicitly * Change force flag to minimum_backup_interval * Don't attempt to open folders on Windows * Join last backup thread before starting new one Also refactor. * Disable auto sync and backup when restoring again * Force backup on full download * Include the reason why a media file import failed, and the file path - Introduce a FileIoError that contains a string representation of the underlying I/O error, and an associated path. There are a few places in the code where we're currently manually including the filename in a custom error message, and this is a step towards a more consistent approach (but we may be better served with a more general approach in the future similar to Anyhow's .context()) - Move the error message into importing.ftl, as it's a bit neater when error messages live in the same file as the rest of the messages associated with some functionality. * Fix importing of media files * Minor wording tweaks * Save an allocation I18n strings with replacements are already strings, so we can skip the extra allocation. Not that it matters here at all. * Terminate import if file missing from archive If a third-party tool is creating invalid archives, the user should know about it. This should be rare, so I did not attempt to make it translatable. * Skip multithreaded compression on small collections Co-authored-by: Damien Elmes <gpg@ankiweb.net>
2022-03-07 06:11:31 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "936cfd212a0155903bcbc060e316fb6cc7cbf2e1907329391ebadc1fe0ce77c2"
Backups (#1685) * Add zstd dep * Implement backend backup with zstd * Implement backup thinning * Write backup meta * Use new file ending anki21b * Asynchronously backup on collection close in Rust * Revert "Add zstd dep" This reverts commit 3fcb2141d2be15f907269d13275c41971431385c. * Add zstd again * Take backup col path from col struct * Fix formatting * Implement backup restoring on backend * Normalize restored media file names * Refactor `extract_legacy_data()` A bit cumbersome due to borrowing rules. * Refactor * Make thinning calendar-based and gradual * Consider last kept backups of previous stages * Import full apkgs and colpkgs with backend * Expose new backup settings * Test `BackupThinner` and make it deterministic * Mark backup_path when closing optional * Delete leaky timer * Add progress updates for restoring media * Write restored collection to tempfile first * Do collection compression in the background thread This has us currently storing an uncompressed and compressed copy of the collection in memory (not ideal), but means the collection can be closed without waiting for compression to complete. On a large collection, this takes a close and reopen from about 0.55s to about 0.07s. The old backup code for comparison: about 0.35s for compression off, about 8.5s for zip compression. * Use multithreading in zstd compression On my system, this reduces the compression time of a large collection from about 0.55s to 0.08s. * Stream compressed collection data into zip file * Tweak backup explanation + Fix incorrect tab order for ignore accents option * Decouple restoring backup and full import In the first case, no profile is opened, unless the new collection succeeds to load. In the second case, either the old collection is reloaded or the new one is loaded. * Fix number gap in Progress message * Don't revert backup when media fails but report it * Tweak error flow * Remove native BackupLimits enum * Fix type annotation * Add thinning test for whole year * Satisfy linter * Await async backup to finish * Move restart disclaimer out of backup tab Should be visible regardless of the current tab. * Write restored collection in chunks * Refactor * Write media in chunks and refactor * Log error if removing file fails * join_backup_task -> await_backup_completion * Refactor backup.rs * Refactor backup meta and collection extraction * Fix wrong error being returned * Call sync_all() on new collection * Add ImportError * Store logger in Backend, instead of creating one on demand init_backend() accepts a Logger rather than a log file, to allow other callers to customize the logger if they wish. In the future we may want to explore using the tracing crate as an alternative; it's a bit more ergonomic, as a logger doesn't need to be passed around, and it plays more nicely with async code. * Sync file contents prior to rename; sync folder after rename. * Limit backup creation to once per 30 min * Use zstd::stream::copy_decode * Make importing abortable * Don't revert if backup media is aborted * Set throttle implicitly * Change force flag to minimum_backup_interval * Don't attempt to open folders on Windows * Join last backup thread before starting new one Also refactor. * Disable auto sync and backup when restoring again * Force backup on full download * Include the reason why a media file import failed, and the file path - Introduce a FileIoError that contains a string representation of the underlying I/O error, and an associated path. There are a few places in the code where we're currently manually including the filename in a custom error message, and this is a step towards a more consistent approach (but we may be better served with a more general approach in the future similar to Anyhow's .context()) - Move the error message into importing.ftl, as it's a bit neater when error messages live in the same file as the rest of the messages associated with some functionality. * Fix importing of media files * Minor wording tweaks * Save an allocation I18n strings with replacements are already strings, so we can skip the extra allocation. Not that it matters here at all. * Terminate import if file missing from archive If a third-party tool is creating invalid archives, the user should know about it. This should be rare, so I did not attempt to make it translatable. * Skip multithreaded compression on small collections Co-authored-by: Damien Elmes <gpg@ankiweb.net>
2022-03-07 06:11:31 +01:00
dependencies = [
"libc",
]
[[package]]
name = "js-sys"
version = "0.3.61"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "445dde2150c55e483f3d8416706b97ec8e8237c307e5b7b4b8dd15e6af2a0730"
dependencies = [
"wasm-bindgen",
]
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
[[package]]
name = "junction"
version = "1.0.0"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "ca39ef0d69b18e6a2fd14c2f0a1d593200f4a4ed949b240b5917ab51fac754cb"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
dependencies = [
"scopeguard",
"winapi",
]
[[package]]
name = "kqueue"
version = "1.0.7"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "2c8fc60ba15bf51257aa9807a48a61013db043fcf3a78cb0d916e8e396dcad98"
dependencies = [
"kqueue-sys",
"libc",
]
[[package]]
name = "kqueue-sys"
version = "1.0.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "8367585489f01bc55dd27404dcf56b95e6da061a256a666ab23be9ba96a2e587"
dependencies = [
"bitflags 1.3.2",
"libc",
]
[[package]]
name = "lazy_static"
version = "1.4.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "e2abad23fbc42b3700f2f279844dc832adb2b2eb069b2df918f455c4e18cc646"
[[package]]
name = "libc"
version = "0.2.146"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "f92be4933c13fd498862a9e02a3055f8a8d9c039ce33db97306fd5a6caa7f29b"
[[package]]
name = "libsqlite3-sys"
version = "0.26.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "afc22eff61b133b115c6e8c74e818c628d6d5e7a502afea6f64dee076dd94326"
dependencies = [
"cc",
"pkg-config",
"vcpkg",
]
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
[[package]]
name = "line-wrap"
version = "0.1.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "f30344350a2a51da54c1d53be93fade8a237e545dbcc4bdbe635413f2117cab9"
dependencies = [
"safemem",
]
[[package]]
name = "link-cplusplus"
version = "1.0.8"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "ecd207c9c713c34f95a097a5b029ac2ce6010530c7b49d7fea24d977dede04f5"
dependencies = [
"cc",
]
[[package]]
name = "linkcheck"
version = "0.4.1"
source = "git+https://github.com/ankitects/linkcheck.git?rev=184b2ca50ed39ca43da13f0b830a463861adb9ca#184b2ca50ed39ca43da13f0b830a463861adb9ca"
dependencies = [
"bytes",
"codespan",
"dunce",
"futures",
"html5ever",
"http",
"lazy_static",
"linkify",
"log",
"markup5ever_rcdom",
"mdbook",
2022-09-24 04:39:21 +02:00
"pulldown-cmark 0.8.0",
"regex",
2021-10-02 12:42:03 +02:00
"reqwest",
"serde",
"thiserror",
"url",
]
[[package]]
name = "linkchecker"
version = "0.0.0"
dependencies = [
"anki",
"futures",
"itertools",
"lazy_static",
"linkcheck",
"regex",
"reqwest",
"strum",
"tokio",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
"workspace-hack",
]
[[package]]
name = "linkify"
version = "0.7.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "04d828fdc1ffceb369a5a9183bd4df2dbb3678f40c8b3fbaa9231de32beb29f9"
dependencies = [
"memchr",
]
[[package]]
name = "linux-raw-sys"
version = "0.3.8"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "ef53942eb7bf7ff43a617b3e2c1c4a5ecf5944a7c1bc12d7ee39bbb15e5c1519"
[[package]]
name = "lock_api"
2022-09-24 04:39:21 +02:00
version = "0.4.9"
source = "registry+https://github.com/rust-lang/crates.io-index"
2022-09-24 04:39:21 +02:00
checksum = "435011366fe56583b16cf956f9df0095b405b82d76425bc8981c0e22e60ec4df"
dependencies = [
2022-09-24 04:39:21 +02:00
"autocfg",
"scopeguard",
]
[[package]]
name = "log"
2022-09-24 04:39:21 +02:00
version = "0.4.17"
source = "registry+https://github.com/rust-lang/crates.io-index"
2022-09-24 04:39:21 +02:00
checksum = "abb12e687cfb44aa40f41fc3978ef76448f9b6038cad6aef4259d3c095a2382e"
dependencies = [
2021-10-02 12:42:03 +02:00
"cfg-if",
]
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
[[package]]
name = "lzma-sys"
version = "0.1.20"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "5fda04ab3764e6cde78b9974eec4f779acaba7c4e84b36eca3cf77c581b85d27"
dependencies = [
"cc",
"libc",
"pkg-config",
]
[[package]]
name = "mac"
version = "0.1.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "c41e0c4fef86961ac6d6f8a82609f55f31b05e4fce149ac5710e439df7619ba4"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
[[package]]
name = "makeapp"
version = "0.0.0"
dependencies = [
"anyhow",
"apple-bundles",
"camino",
"clap 4.2.1",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
"glob",
"lazy_static",
"plist",
"serde",
"serde_json",
2022-11-30 03:32:27 +01:00
"simple-file-manifest",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
"walkdir",
"workspace-hack",
]
[[package]]
name = "makeinstall"
version = "0.0.0"
dependencies = [
"anyhow",
"camino",
"clap 4.2.1",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
"glob",
"tugger-windows-codesign",
"walkdir",
"workspace-hack",
]
[[package]]
name = "maplit"
version = "1.0.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "3e2e65a1a2e43cfcb47a895c4c8b10d1f4a61097f9f254f183aee60cad9c651d"
[[package]]
name = "markup5ever"
version = "0.11.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "7a2629bb1404f3d34c2e921f21fd34ba00b206124c81f65c50b43b6aaefeb016"
dependencies = [
"log",
"phf 0.10.1",
"phf_codegen",
"string_cache",
"string_cache_codegen",
"tendril",
]
[[package]]
name = "markup5ever_rcdom"
version = "0.2.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "b9521dd6750f8e80ee6c53d65e2e4656d7de37064f3a7a5d2d11d05df93839c2"
dependencies = [
"html5ever",
"markup5ever",
"tendril",
"xml5ever",
]
[[package]]
name = "matchers"
version = "0.1.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "8263075bb86c5a1b1427b5ae862e8889656f126e9f77c484496e8b47cf5c5558"
dependencies = [
"regex-automata",
]
[[package]]
name = "matches"
version = "0.1.10"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "2532096657941c2fea9c289d370a250971c689d4f143798ff67113ec042024a5"
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
[[package]]
name = "matchit"
version = "0.7.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "b87248edafb776e59e6ee64a79086f65890d3510f2c656c000bf2a7e8a0aea40"
[[package]]
name = "mdbook"
version = "0.4.28"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "764dcbfc2e5f868bc1b566eb179dff1a06458fd0cff846aae2579392dd3f01a0"
dependencies = [
"ammonia",
"anyhow",
"chrono",
"clap 4.2.1",
"clap_complete",
"elasticlunr-rs",
"env_logger",
"futures-util",
"gitignore",
"handlebars",
"log",
"memchr",
"notify",
"notify-debouncer-mini",
"once_cell",
"opener",
"pulldown-cmark 0.9.2",
"regex",
"serde",
"serde_json",
"shlex",
"tempfile",
"tokio",
"toml",
"topological-sort",
"warp",
]
[[package]]
name = "memchr"
2022-09-24 04:39:21 +02:00
version = "2.5.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
2022-09-24 04:39:21 +02:00
checksum = "2dffe52ecf27772e601905b7522cb4ef790d2cc203488bbd0e2fe85fcb74566d"
[[package]]
name = "memoffset"
version = "0.8.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "d61c719bcfbcf5d62b3a09efa6088de8c54bc0bfcd3ea7ae39fcc186108b8de1"
dependencies = [
"autocfg",
]
[[package]]
name = "mime"
version = "0.3.17"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "6877bb514081ee2a7ff5ef9de3281f14a4dd4bceac4c09388074a6b5df8a139a"
[[package]]
name = "mime_guess"
version = "2.0.4"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "4192263c238a5f0d0c6bfd21f336a313a4ce1c450542449ca191bb657b4642ef"
dependencies = [
"mime",
"unicase",
]
[[package]]
name = "minilints"
version = "0.0.0"
dependencies = [
"anyhow",
"camino",
"once_cell",
"walkdir",
]
2021-10-02 12:42:03 +02:00
[[package]]
name = "minimal-lexical"
2021-11-18 11:54:00 +01:00
version = "0.2.1"
2021-10-02 12:42:03 +02:00
source = "registry+https://github.com/rust-lang/crates.io-index"
2021-11-18 11:54:00 +01:00
checksum = "68354c5c6bd36d73ff3feceb05efa59b6acb7626617f4962be322a825e61f79a"
2021-10-02 12:42:03 +02:00
[[package]]
name = "miniz_oxide"
version = "0.6.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "b275950c28b37e794e8c55d88aeb5e139d0ce23fdbbeda68f8d7174abdf9e8fa"
dependencies = [
"adler",
]
[[package]]
name = "mio"
version = "0.8.6"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "5b9d9a46eff5b4ff64b45a9e316a6d1e0bc719ef429cbec4dc630684212bfdf9"
dependencies = [
"libc",
"log",
"wasi 0.11.0+wasi-snapshot-preview1",
"windows-sys 0.45.0",
]
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
[[package]]
name = "multer"
version = "2.0.4"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "6ed4198ce7a4cbd2a57af78d28c6fbb57d81ac5f1d6ad79ac6c5587419cbdf22"
dependencies = [
"bytes",
"encoding_rs",
"futures-util",
"http",
"httparse",
"log",
"memchr",
"mime",
"spin 0.9.8",
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
"version_check",
]
[[package]]
name = "multimap"
2021-03-27 05:47:16 +01:00
version = "0.8.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
2021-03-27 05:47:16 +01:00
checksum = "e5ce46fe64a9d73be07dcbe690a38ce1b293be448fd8ce1e6c1b8062c9f72c6a"
[[package]]
name = "native-tls"
version = "0.2.11"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "07226173c32f2926027b63cce4bcd8076c3552846cbe7925f3aaffeac0a3b92e"
dependencies = [
"lazy_static",
"libc",
"log",
"openssl",
"openssl-probe",
"openssl-sys",
"schannel",
"security-framework",
"security-framework-sys",
"tempfile",
]
[[package]]
name = "new_debug_unreachable"
version = "1.0.4"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "e4a24736216ec316047a1fc4252e27dabb04218aa4a3f37c6e7ddbf1f9782b54"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
[[package]]
name = "ninja_gen"
version = "0.0.0"
dependencies = [
"anki_io",
"anyhow",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
"camino",
"globset",
"itertools",
"lazy_static",
"maplit",
"num_cpus",
"walkdir",
"which",
"workspace-hack",
]
2021-06-25 07:35:25 +02:00
[[package]]
name = "nom"
version = "7.1.3"
2021-06-25 07:35:25 +02:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "d273983c5a657a70a3e8f2a01329822f3b8c8172b73826411a55751e404a0a4a"
2021-06-25 07:35:25 +02:00
dependencies = [
"memchr",
2021-10-02 12:42:03 +02:00
"minimal-lexical",
2021-06-25 07:35:25 +02:00
]
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
[[package]]
name = "nonempty"
version = "0.7.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "e9e591e719385e6ebaeb5ce5d3887f7d5676fceca6411d1925ccc95745f3d6f7"
[[package]]
name = "notify"
version = "5.1.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "58ea850aa68a06e48fdb069c0ec44d0d64c8dbffa49bf3b6f7f0a901fdea1ba9"
dependencies = [
"bitflags 1.3.2",
"crossbeam-channel",
"filetime",
"fsevent-sys",
"inotify",
"kqueue",
"libc",
"mio",
"walkdir",
"windows-sys 0.42.0",
]
[[package]]
name = "notify-debouncer-mini"
version = "0.2.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "e23e9fa24f094b143c1eb61f90ac6457de87be6987bc70746e0179f7dbc9007b"
dependencies = [
"crossbeam-channel",
"notify",
]
[[package]]
name = "nu-ansi-term"
version = "0.46.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "77a8165726e8236064dbb45459242600304b42a5ea24ee2948e18e023bf7ba84"
dependencies = [
"overload",
"winapi",
]
[[package]]
name = "num-format"
version = "0.4.4"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "a652d9771a63711fd3c3deb670acfbe5c30a4072e664d7a3bf5a9e1056ac72c3"
dependencies = [
"arrayvec",
"itoa",
]
[[package]]
name = "num-integer"
2022-09-24 04:39:21 +02:00
version = "0.1.45"
source = "registry+https://github.com/rust-lang/crates.io-index"
2022-09-24 04:39:21 +02:00
checksum = "225d3389fb3509a24c93f5c29eb6bde2586b98d9f016636dff58d7c6f7569cd9"
dependencies = [
"autocfg",
"num-traits",
]
[[package]]
name = "num-traits"
2022-09-24 04:39:21 +02:00
version = "0.2.15"
source = "registry+https://github.com/rust-lang/crates.io-index"
2022-09-24 04:39:21 +02:00
checksum = "578ede34cf02f8924ab9447f50c28075b4d3e5b269972345e7e0372b38c6cdcd"
dependencies = [
"autocfg",
]
[[package]]
name = "num_cpus"
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
version = "1.15.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
checksum = "0fac9e2da13b5eb447a6ce3d392f23a29d8694bff781bf03a16cd9ac8697593b"
dependencies = [
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
"hermit-abi 0.2.6",
"libc",
]
[[package]]
name = "num_enum"
version = "0.6.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "7a015b430d3c108a207fd776d2e2196aaf8b1cf8cf93253e3a097ff3085076a1"
dependencies = [
"num_enum_derive",
]
[[package]]
name = "num_enum_derive"
version = "0.6.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "96667db765a921f7b295ffee8b60472b686a51d4f21c2ee4ffdb94c7013b65a6"
dependencies = [
"proc-macro-crate",
"proc-macro2",
"quote",
"syn 2.0.12",
]
[[package]]
name = "object"
version = "0.30.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "ea86265d3d3dcb6a27fc51bd29a4bf387fae9d2986b823079d4986af253eb439"
dependencies = [
"memchr",
]
[[package]]
name = "once_cell"
version = "1.17.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "b7e5500299e16ebb147ae15a00a942af264cf3688f47923b8fc2cd5858f23ad3"
[[package]]
name = "oorandom"
version = "11.1.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "0ab1bc2a289d34bd04a330323ac98a1b4bc82c9d9fcb1e66b63caa84da26b575"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
[[package]]
name = "opaque-debug"
version = "0.3.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "624a8340c38c1b80fd549087862da4ba43e08858af025b236e509b6649fc13d5"
[[package]]
name = "opener"
version = "0.5.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "293c15678e37254c15bd2f092314abb4e51d7fdde05c2021279c12631b54f005"
dependencies = [
"bstr 1.4.0",
"winapi",
]
[[package]]
name = "openssl"
version = "0.10.48"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "518915b97df115dd36109bfa429a48b8f737bd05508cf9588977b599648926d2"
dependencies = [
"bitflags 1.3.2",
2021-10-02 12:42:03 +02:00
"cfg-if",
"foreign-types",
"libc",
2021-03-27 05:47:16 +01:00
"once_cell",
2022-09-24 04:39:21 +02:00
"openssl-macros",
"openssl-sys",
]
2022-09-24 04:39:21 +02:00
[[package]]
name = "openssl-macros"
version = "0.1.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "b501e44f11665960c7e7fcf062c7d96a14ade4aa98116c004b2e37b5be7d736c"
dependencies = [
"proc-macro2",
"quote",
"syn 1.0.109",
2022-09-24 04:39:21 +02:00
]
2020-11-03 01:26:27 +01:00
[[package]]
name = "openssl-probe"
2022-01-15 05:59:43 +01:00
version = "0.1.5"
2020-11-03 01:26:27 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
2022-01-15 05:59:43 +01:00
checksum = "ff011a302c396a5197692431fc1948019154afc178baf7d8e37367442a4601cf"
2020-11-03 01:26:27 +01:00
[[package]]
name = "openssl-sys"
version = "0.9.83"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "666416d899cf077260dac8698d60a60b435a46d57e82acb1be3d0dad87284e5b"
dependencies = [
"autocfg",
"cc",
"libc",
"pkg-config",
"vcpkg",
]
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
[[package]]
name = "os_pipe"
version = "1.1.3"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "a53dbb20faf34b16087a931834cba2d7a73cc74af2b7ef345a4c8324e2409a12"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
dependencies = [
"libc",
"windows-sys 0.45.0",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
]
[[package]]
2022-09-24 04:39:21 +02:00
name = "os_str_bytes"
version = "6.5.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "ceedf44fb00f2d1984b0bc98102627ce622e083e49a5bacdb3e514fa4238e267"
[[package]]
name = "overload"
version = "0.1.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "b15813163c1d831bf4a13c3610c05c0d03b39feb07f7e09fa234dac9b15aaf39"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
[[package]]
name = "p12"
2022-11-30 03:32:27 +01:00
version = "0.6.3"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
2022-11-30 03:32:27 +01:00
checksum = "d4873306de53fe82e7e484df31e1e947d61514b6ea2ed6cd7b45d63006fd9224"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
dependencies = [
2022-11-30 03:32:27 +01:00
"cbc",
"cipher 0.4.4",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
"des",
"getrandom 0.2.8",
2022-11-30 03:32:27 +01:00
"hmac",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
"lazy_static",
"rc2",
2022-11-30 03:32:27 +01:00
"sha1",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
"yasna",
]
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
[[package]]
name = "parking"
version = "2.0.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "427c3892f9e783d91cc128285287e70a59e206ca452770ece88a76f7a3eddd72"
[[package]]
name = "parking_lot"
2022-09-24 04:39:21 +02:00
version = "0.12.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
2022-09-24 04:39:21 +02:00
checksum = "3742b2c103b9f06bc9fff0a37ff4912935851bee6d36f3c02bcc755bcfec228f"
dependencies = [
"lock_api",
2022-09-24 04:39:21 +02:00
"parking_lot_core",
]
[[package]]
name = "parking_lot_core"
version = "0.9.7"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "9069cbb9f99e3a5083476ccb29ceb1de18b9118cafa53e90c9551235de2b9521"
dependencies = [
"cfg-if",
"libc",
"redox_syscall 0.2.16",
"smallvec",
"windows-sys 0.45.0",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
]
[[package]]
name = "password-hash"
version = "0.4.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "7676374caaee8a325c9e7a2ae557f216c5563a171d6997b0ef8a65af35147700"
dependencies = [
"base64ct",
"rand_core 0.6.4",
"subtle",
]
[[package]]
name = "pbkdf2"
version = "0.11.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "83a0692ec44e4cf1ef28ca317f14f8f07da2d95ec3fa01f86e4467b725e60917"
dependencies = [
2022-11-30 03:32:27 +01:00
"digest",
"hmac",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
"password-hash",
2022-11-30 03:32:27 +01:00
"sha2",
]
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
[[package]]
name = "pem"
version = "1.1.1"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "a8835c273a76a90455d7344889b0964598e3316e2a79ede8e36f16bdcf2228b8"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
dependencies = [
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
"base64 0.13.1",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
]
[[package]]
name = "percent-encoding"
2022-09-24 04:39:21 +02:00
version = "2.2.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
2022-09-24 04:39:21 +02:00
checksum = "478c572c3d73181ff3c2539045f6eb99e5491218eae919370993b890cdbdd98e"
[[package]]
name = "percent-encoding-iri"
version = "2.2.0"
source = "git+https://github.com/ankitects/rust-url.git?rev=bb930b8d089f4d30d7d19c12e54e66191de47b88#bb930b8d089f4d30d7d19c12e54e66191de47b88"
[[package]]
name = "pest"
version = "2.5.7"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "7b1403e8401ad5dedea73c626b99758535b342502f8d1e361f4a2dd952749122"
dependencies = [
"thiserror",
"ucd-trie",
]
[[package]]
name = "pest_derive"
version = "2.5.7"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "be99c4c1d2fc2769b1d00239431d711d08f6efedcecb8b6e30707160aee99c15"
2021-06-25 07:35:25 +02:00
dependencies = [
"pest",
"pest_generator",
]
[[package]]
name = "pest_generator"
version = "2.5.7"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "e56094789873daa36164de2e822b3888c6ae4b4f9da555a1103587658c805b1e"
dependencies = [
"pest",
"pest_meta",
"proc-macro2",
"quote",
"syn 2.0.12",
]
[[package]]
name = "pest_meta"
version = "2.5.7"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "6733073c7cff3d8459fda0e42f13a047870242aed8b509fe98000928975f359e"
dependencies = [
"once_cell",
"pest",
"sha2",
]
[[package]]
name = "petgraph"
version = "0.6.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "4dd7d28ee937e54fe3080c91faa1c3a46c06de6252988a7f4592ba2310ef22a4"
dependencies = [
"fixedbitset",
"indexmap",
2021-06-25 07:35:25 +02:00
]
[[package]]
name = "phf"
2022-01-15 05:59:43 +01:00
version = "0.10.1"
2021-06-25 07:35:25 +02:00
source = "registry+https://github.com/rust-lang/crates.io-index"
2022-01-15 05:59:43 +01:00
checksum = "fabbf1ead8a5bcbc20f5f8b939ee3f5b0f6f281b6ad3468b84656b658b455259"
dependencies = [
2021-10-02 12:42:03 +02:00
"phf_shared 0.10.0",
2022-09-24 04:39:21 +02:00
]
[[package]]
name = "phf"
version = "0.11.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "928c6535de93548188ef63bb7c4036bd415cd8f36ad25af44b9789b2ee72a48c"
dependencies = [
"phf_macros",
2022-09-24 04:39:21 +02:00
"phf_shared 0.11.1",
]
2022-09-24 04:39:21 +02:00
[[package]]
name = "phf_codegen"
version = "0.10.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "4fb1c3a8bc4dd4e5cfce29b44ffc14bedd2ee294559a294e2a4d4c9e9a6a13cd"
dependencies = [
"phf_generator 0.10.0",
"phf_shared 0.10.0",
]
2021-06-25 07:35:25 +02:00
[[package]]
name = "phf_generator"
2021-10-02 12:42:03 +02:00
version = "0.10.0"
2021-06-25 07:35:25 +02:00
source = "registry+https://github.com/rust-lang/crates.io-index"
2021-10-02 12:42:03 +02:00
checksum = "5d5285893bb5eb82e6aaf5d59ee909a06a16737a8970984dd7746ba9283498d6"
2021-06-25 07:35:25 +02:00
dependencies = [
2021-10-02 12:42:03 +02:00
"phf_shared 0.10.0",
"rand 0.8.5",
2021-06-25 07:35:25 +02:00
]
2022-09-24 04:39:21 +02:00
[[package]]
name = "phf_generator"
version = "0.11.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "b1181c94580fa345f50f19d738aaa39c0ed30a600d95cb2d3e23f94266f14fbf"
dependencies = [
"phf_shared 0.11.1",
"rand 0.8.5",
]
[[package]]
name = "phf_macros"
2022-09-24 04:39:21 +02:00
version = "0.11.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
2022-09-24 04:39:21 +02:00
checksum = "92aacdc5f16768709a569e913f7451034034178b05bdc8acda226659a3dccc66"
dependencies = [
2022-09-24 04:39:21 +02:00
"phf_generator 0.11.1",
"phf_shared 0.11.1",
"proc-macro2",
"quote",
"syn 1.0.109",
]
2021-06-25 07:35:25 +02:00
[[package]]
name = "phf_shared"
2021-10-02 12:42:03 +02:00
version = "0.10.0"
2021-06-25 07:35:25 +02:00
source = "registry+https://github.com/rust-lang/crates.io-index"
2021-10-02 12:42:03 +02:00
checksum = "b6796ad771acdc0123d2a88dc428b5e38ef24456743ddb1744ed628f9815c096"
2021-06-25 07:35:25 +02:00
dependencies = [
"siphasher",
]
2022-09-24 04:39:21 +02:00
[[package]]
name = "phf_shared"
version = "0.11.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "e1fb5f6f826b772a8d4c0394209441e7d37cbbb967ae9c7e0e8134365c9ee676"
dependencies = [
"siphasher",
]
[[package]]
name = "pin-project"
2022-09-24 04:39:21 +02:00
version = "1.0.12"
source = "registry+https://github.com/rust-lang/crates.io-index"
2022-09-24 04:39:21 +02:00
checksum = "ad29a609b6bcd67fee905812e544992d216af9d755757c05ed2d0e15a74c6ecc"
dependencies = [
2021-03-07 10:04:34 +01:00
"pin-project-internal",
]
[[package]]
name = "pin-project-internal"
2022-09-24 04:39:21 +02:00
version = "1.0.12"
source = "registry+https://github.com/rust-lang/crates.io-index"
2022-09-24 04:39:21 +02:00
checksum = "069bdb1e05adc7a8990dce9cc75370895fbe4e3d58b9b73bf1aee56359344a55"
dependencies = [
"proc-macro2",
"quote",
"syn 1.0.109",
]
2020-11-24 07:57:37 +01:00
[[package]]
name = "pin-project-lite"
2022-09-24 04:39:21 +02:00
version = "0.2.9"
2020-11-24 07:57:37 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
2022-09-24 04:39:21 +02:00
checksum = "e0a7ae3ac2f1173085d398531c705756c94a4c56843785df85a60c1a0afac116"
2020-11-24 07:57:37 +01:00
[[package]]
name = "pin-utils"
version = "0.1.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "8b870d8c151b6f2fb93e84a13146138f05d02ed11c7e7c54f8826aaaf7c9f184"
[[package]]
name = "pkg-config"
version = "0.3.26"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "6ac9a59f73473f1b8d852421e59e64809f025994837ef743615c6d0c5b305160"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
[[package]]
name = "plist"
version = "1.4.3"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "9bd9647b268a3d3e14ff09c23201133a62589c658db02bb7388c7246aafe0590"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
dependencies = [
"base64 0.21.0",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
"indexmap",
"line-wrap",
"quick-xml",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
"serde",
"time",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
]
[[package]]
name = "plotters"
2022-09-24 04:39:21 +02:00
version = "0.3.4"
source = "registry+https://github.com/rust-lang/crates.io-index"
2022-09-24 04:39:21 +02:00
checksum = "2538b639e642295546c50fcd545198c9d64ee2a38620a628724a3b266d5fbf97"
dependencies = [
"num-traits",
"plotters-backend",
"plotters-svg",
"wasm-bindgen",
"web-sys",
]
[[package]]
name = "plotters-backend"
2022-09-24 04:39:21 +02:00
version = "0.3.4"
source = "registry+https://github.com/rust-lang/crates.io-index"
2022-09-24 04:39:21 +02:00
checksum = "193228616381fecdc1224c62e96946dfbc73ff4384fba576e052ff8c1bea8142"
[[package]]
name = "plotters-svg"
2022-09-24 04:39:21 +02:00
version = "0.3.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
2022-09-24 04:39:21 +02:00
checksum = "f9a81d2759aae1dae668f783c308bc5c8ebd191ff4184aaa1b37f65a6ae5a56f"
dependencies = [
"plotters-backend",
]
[[package]]
name = "ppv-lite86"
version = "0.2.17"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "5b40af805b3121feab8a3c29f04d8ad262fa8e0561883e7653e024ae4479e6de"
[[package]]
name = "precomputed-hash"
version = "0.1.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "925383efa346730478fb4838dbe9137d2a47675ad789c546d150a6e1dd4ab31c"
[[package]]
name = "prettyplease"
version = "0.1.25"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "6c8646e95016a7a6c4adea95bafa8a16baab64b583356217f2c85db4a39d9a86"
dependencies = [
"proc-macro2",
"syn 1.0.109",
]
[[package]]
name = "proc-macro-crate"
version = "1.3.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "7f4c021e1093a56626774e81216a4ce732a735e5bad4868a03f3ed65ca0c3919"
dependencies = [
2022-09-24 04:39:21 +02:00
"once_cell",
"toml_edit",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
]
[[package]]
name = "proc-macro-hack"
version = "0.5.20+deprecated"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "dc375e1527247fe1a97d8b7156678dfe7c1af2fc075c9a4db3690ecd2a148068"
[[package]]
name = "proc-macro2"
version = "1.0.54"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "e472a104799c74b514a57226160104aa483546de37e839ec50e3c2e41dd87534"
dependencies = [
2022-09-24 04:39:21 +02:00
"unicode-ident",
]
[[package]]
name = "prost"
version = "0.11.9"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "0b82eaa1d779e9a4bc1c3217db8ffbeabaae1dca241bf70183242128d48681cd"
dependencies = [
"bytes",
"prost-derive",
]
[[package]]
name = "prost-build"
version = "0.11.9"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "119533552c9a7ffacc21e099c24a0ac8bb19c2a2a3f363de84cd9b844feab270"
dependencies = [
"bytes",
"heck",
2021-10-02 12:42:03 +02:00
"itertools",
2021-11-18 11:54:00 +01:00
"lazy_static",
"log",
"multimap",
"petgraph",
"prettyplease",
"prost",
"prost-types",
2021-11-18 11:54:00 +01:00
"regex",
"syn 1.0.109",
"tempfile",
"which",
]
[[package]]
name = "prost-derive"
version = "0.11.9"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "e5d2d8d10f3c6ded6da8b05b5fb3b8a5082514344d56c9f871412d29b4e075b4"
dependencies = [
"anyhow",
2021-10-02 12:42:03 +02:00
"itertools",
"proc-macro2",
"quote",
"syn 1.0.109",
]
[[package]]
name = "prost-reflect"
version = "0.11.4"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "000e1e05ebf7b26e1eba298e66fe4eee6eb19c567d0ffb35e0dd34231cdac4c8"
dependencies = [
"once_cell",
"prost",
"prost-types",
]
[[package]]
name = "prost-types"
version = "0.11.9"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "213622a1460818959ac1181aaeb2dc9c7f63df720db7d788b3e24eacd1983e13"
dependencies = [
"prost",
]
[[package]]
name = "pulldown-cmark"
version = "0.8.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "ffade02495f22453cd593159ea2f59827aae7f53fa8323f756799b670881dcf8"
dependencies = [
"bitflags 1.3.2",
"getopts",
"memchr",
"unicase",
]
2022-09-24 04:39:21 +02:00
[[package]]
name = "pulldown-cmark"
version = "0.9.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "2d9cc634bc78768157b5cbfe988ffcd1dcba95cd2b2f03a88316c08c6d00ed63"
dependencies = [
"bitflags 1.3.2",
2022-09-24 04:39:21 +02:00
"getopts",
"memchr",
"unicase",
]
[[package]]
name = "pyo3"
version = "0.18.2"
2020-12-23 01:08:38 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "cfb848f80438f926a9ebddf0a539ed6065434fd7aae03a89312a9821f81b8501"
dependencies = [
2021-10-02 12:42:03 +02:00
"cfg-if",
"indoc",
"libc",
"memoffset",
2022-09-24 04:39:21 +02:00
"parking_lot",
2021-10-02 12:42:03 +02:00
"pyo3-build-config",
2022-09-24 04:39:21 +02:00
"pyo3-ffi",
2020-12-23 01:08:38 +01:00
"pyo3-macros",
"unindent",
]
2021-10-02 12:42:03 +02:00
[[package]]
name = "pyo3-build-config"
version = "0.18.2"
2021-10-02 12:42:03 +02:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "98a42e7f42e917ce6664c832d5eee481ad514c98250c49e0b03b20593e2c7ed0"
2021-10-02 12:42:03 +02:00
dependencies = [
"once_cell",
2022-09-24 04:39:21 +02:00
"target-lexicon",
]
[[package]]
name = "pyo3-ffi"
version = "0.18.2"
2022-09-24 04:39:21 +02:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "a0707f0ab26826fe4ccd59b69106e9df5e12d097457c7b8f9c0fd1d2743eec4d"
2022-09-24 04:39:21 +02:00
dependencies = [
"libc",
"pyo3-build-config",
2021-10-02 12:42:03 +02:00
]
[[package]]
2020-12-23 01:08:38 +01:00
name = "pyo3-macros"
version = "0.18.2"
2020-12-23 01:08:38 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "978d18e61465ecd389e1f235ff5a467146dc4e3c3968b90d274fe73a5dd4a438"
dependencies = [
2022-09-24 04:39:21 +02:00
"proc-macro2",
2020-12-23 01:08:38 +01:00
"pyo3-macros-backend",
"quote",
"syn 1.0.109",
]
[[package]]
2020-12-23 01:08:38 +01:00
name = "pyo3-macros-backend"
version = "0.18.2"
2020-12-23 01:08:38 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "8e0e1128f85ce3fca66e435e08aa2089a2689c1c48ce97803e13f63124058462"
dependencies = [
2020-12-23 01:08:38 +01:00
"proc-macro2",
"quote",
"syn 1.0.109",
]
[[package]]
name = "quick-xml"
version = "0.28.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "e5c1a97b1bc42b1d550bfb48d4262153fe400a12bab1511821736f7eac76d7e2"
dependencies = [
"memchr",
]
[[package]]
name = "quote"
version = "1.0.26"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "4424af4bf778aae2051a77b60283332f386554255d722233d09fbfc7e30da2fc"
dependencies = [
"proc-macro2",
]
[[package]]
name = "rand"
version = "0.7.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "6a6b1679d49b24bbfe0c803429aa1874472f50d9b363131f0e89fc356b544d03"
dependencies = [
2021-02-03 11:29:48 +01:00
"getrandom 0.1.16",
"libc",
2021-02-03 11:29:48 +01:00
"rand_chacha 0.2.2",
"rand_core 0.5.1",
"rand_hc",
2021-02-03 11:29:48 +01:00
]
[[package]]
name = "rand"
version = "0.8.5"
2021-02-03 11:29:48 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "34af8d1a0e25924bc5b7c43c079c942339d8f0a8b57c39049bef581b46327404"
2021-02-03 11:29:48 +01:00
dependencies = [
"libc",
2021-06-16 08:10:57 +02:00
"rand_chacha 0.3.1",
2022-09-24 04:39:21 +02:00
"rand_core 0.6.4",
]
[[package]]
name = "rand_chacha"
version = "0.2.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "f4c8ed856279c9737206bf725bf36935d8666ead7aa69b52be55af369d193402"
dependencies = [
"ppv-lite86",
2021-02-03 11:29:48 +01:00
"rand_core 0.5.1",
]
[[package]]
name = "rand_chacha"
2021-06-16 08:10:57 +02:00
version = "0.3.1"
2021-02-03 11:29:48 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
2021-06-16 08:10:57 +02:00
checksum = "e6c10a63a0fa32252be49d21e7709d4d4baf8d231c2dbce1eaa8141b9b127d88"
2021-02-03 11:29:48 +01:00
dependencies = [
"ppv-lite86",
2022-09-24 04:39:21 +02:00
"rand_core 0.6.4",
]
[[package]]
name = "rand_core"
version = "0.5.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "90bde5296fc891b0cef12a6d03ddccc162ce7b2aff54160af9338f8d40df6d19"
dependencies = [
2021-02-03 11:29:48 +01:00
"getrandom 0.1.16",
]
[[package]]
name = "rand_core"
2022-09-24 04:39:21 +02:00
version = "0.6.4"
2021-02-03 11:29:48 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
2022-09-24 04:39:21 +02:00
checksum = "ec0be4795e2f6a28069bec0b5ff3e2ac9bafc99e6a9a7dc3547996c5c816922c"
2021-02-03 11:29:48 +01:00
dependencies = [
"getrandom 0.2.8",
]
[[package]]
name = "rand_hc"
version = "0.2.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "ca3129af7b92a17112d59ad498c6f81eaf463253766b90396d39ea7a39d6613c"
dependencies = [
2021-02-03 11:29:48 +01:00
"rand_core 0.5.1",
]
[[package]]
name = "rayon"
version = "1.7.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "1d2df5196e37bcc87abebc0053e20787d73847bb33134a69841207dd0a47f03b"
dependencies = [
"either",
"rayon-core",
]
[[package]]
name = "rayon-core"
version = "1.11.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "4b8f95bd6966f5c87776639160a66bd8ab9895d9d4ab01ddba9fc60661aebe8d"
dependencies = [
"crossbeam-channel",
"crossbeam-deque",
"crossbeam-utils",
"num_cpus",
]
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
[[package]]
name = "rc2"
2022-11-30 03:32:27 +01:00
version = "0.8.1"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
2022-11-30 03:32:27 +01:00
checksum = "62c64daa8e9438b84aaae55010a93f396f8e60e3911590fcba770d04643fc1dd"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
dependencies = [
"cipher 0.4.4",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
]
[[package]]
name = "rcgen"
version = "0.10.0"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "ffbe84efe2f38dea12e9bfc1f65377fdf03e53a18cb3b995faedf7934c7e785b"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
dependencies = [
"pem",
"ring",
"time",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
"yasna",
]
[[package]]
name = "redox_syscall"
2022-09-24 04:39:21 +02:00
version = "0.2.16"
source = "registry+https://github.com/rust-lang/crates.io-index"
2022-09-24 04:39:21 +02:00
checksum = "fb5a58c1855b4b6819d59012155603f0b22ad30cad752600aadfcb695265519a"
2021-02-03 11:29:48 +01:00
dependencies = [
"bitflags 1.3.2",
]
[[package]]
name = "redox_syscall"
version = "0.3.5"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "567664f262709473930a4bf9e51bf2ebf3348f2e748ccc50dea20646858f8f29"
dependencies = [
"bitflags 1.3.2",
2021-02-03 11:29:48 +01:00
]
[[package]]
name = "regex"
version = "1.7.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "8b1f693b24f6ac912f4893ef08244d70b6067480d2f1a46e950c9691e6749d1d"
dependencies = [
"aho-corasick",
"memchr",
"regex-syntax",
]
[[package]]
name = "regex-automata"
version = "0.1.10"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "6c230d73fb8d8c1b9c0b3135c5142a8acee3a0558fb8db5cf1cb65f8d7862132"
dependencies = [
"regex-syntax",
]
[[package]]
name = "regex-syntax"
version = "0.6.29"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "f162c6dd7b008981e4d40210aca20b4bd0f9b60ca9271061b07f78537722f2e1"
[[package]]
name = "reqwest"
version = "0.11.16"
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "27b71749df584b7f4cac2c426c127a7c785a5106cc98f7a8feb044115f0fa254"
dependencies = [
"base64 0.21.0",
"bytes",
"encoding_rs",
"futures-core",
"futures-util",
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
"h2",
"http",
"http-body",
"hyper",
"hyper-rustls",
"hyper-tls",
2020-11-03 01:26:27 +01:00
"ipnet",
"js-sys",
"log",
"mime",
"mime_guess",
"native-tls",
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
"once_cell",
"percent-encoding",
"pin-project-lite",
"rustls",
2021-06-18 10:15:41 +02:00
"rustls-native-certs",
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
"rustls-pemfile",
"serde",
"serde_json",
"serde_urlencoded",
"tokio",
"tokio-native-tls",
"tokio-rustls",
"tokio-socks",
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
"tokio-util",
"tower-service",
"url",
"wasm-bindgen",
"wasm-bindgen-futures",
"wasm-streams",
"web-sys",
"webpki-roots",
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
"winreg 0.10.1",
]
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
[[package]]
name = "retain_mut"
version = "0.1.9"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "4389f1d5789befaf6029ebd9f7dac4af7f7e3d61b69d4f30e2ac02b57e7712b0"
[[package]]
name = "ring"
2021-02-03 11:29:48 +01:00
version = "0.16.20"
source = "registry+https://github.com/rust-lang/crates.io-index"
2021-02-03 11:29:48 +01:00
checksum = "3053cf52e236a3ed746dfc745aa9cacf1b791d846bdaf412f60a8d7d6e17c8fc"
dependencies = [
"cc",
"libc",
"once_cell",
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
"spin 0.5.2",
"untrusted",
"web-sys",
"winapi",
]
2020-11-24 09:41:03 +01:00
[[package]]
name = "rsbridge"
version = "0.0.0"
dependencies = [
"anki",
"pyo3",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
"workspace-hack",
]
[[package]]
name = "runner"
version = "0.0.0"
dependencies = [
Migrate to protobuf-es (#2547) * Fix .no-reduce-motion missing from graphs spinner, and not being honored * Begin migration from protobuf.js -> protobuf-es Motivation: - Protobuf-es has a nicer API: messages are represented as classes, and fields which should exist are not marked as nullable. - As it uses modules, only the proto messages we actually use get included in our bundle output. Protobuf.js put everything in a namespace, which prevented tree-shaking, and made it awkward to access inner messages. - ./run after touching a proto file drops from about 8s to 6s on my machine. The tradeoff is slower decoding/encoding (#2043), but that was mainly a concern for the graphs page, and was unblocked by https://github.com/ankitects/anki/commit/37151213cd9d431f449ba4b3bc4c0329a1d9af78 Approach/notes: - We generate the new protobuf-es interface in addition to existing protobuf.js interface, so we can migrate a module at a time, starting with the graphs module. - rslib:proto now generates RPC methods for TS in addition to the Python interface. The input-arg-unrolling behaviour of the Python generation is not required here, as we declare the input arg as a PlainMessage<T>, which marks it as requiring all fields to be provided. - i64 is represented as bigint in protobuf-es. We were using a patch to protobuf.js to get it to output Javascript numbers instead of long.js types, but now that our supported browser versions support bigint, it's probably worth biting the bullet and migrating to bigint use. Our IDs fit comfortably within MAX_SAFE_INTEGER, but that may not hold for future fields we add. - Oneofs are handled differently in protobuf-es, and are going to need some refactoring. Other notable changes: - Added a --mkdir arg to our build runner, so we can create a dir easily during the build on Windows. - Simplified the preference handling code, by wrapping the preferences in an outer store, instead of a separate store for each individual preference. This means a change to one preference will trigger a redraw of all components that depend on the preference store, but the redrawing is cheap after moving the data processing to Rust, and it makes the code easier to follow. - Drop async(Reactive).ts in favour of more explicit handling with await blocks/updating. - Renamed add_inputs_to_group() -> add_dependency(), and fixed it not adding dependencies to parent groups. Renamed add() -> add_action() for clarity. * Remove a couple of unused proto imports * Migrate card info * Migrate congrats, image occlusion, and tag editor + Fix imports for multi-word proto files. * Migrate change-notetype * Migrate deck options * Bump target to es2020; simplify ts lib list Have used caniuse.com to confirm Chromium 77, iOS 14.5 and the Chrome on Android support the full es2017-es2020 features. * Migrate import-csv * Migrate i18n and fix missing output types in .js * Migrate custom scheduling, and remove protobuf.js To mostly maintain our old API contract, we make use of protobuf-es's ability to convert to JSON, which follows the same format as protobuf.js did. It doesn't cover all case: users who were previously changing the variant of a type will need to update their code, as assigning to a new variant no longer automatically removes the old one, which will cause an error when we try to convert back from JSON. But I suspect the large majority of users are adjusting the current variant rather than creating a new one, and this saves us having to write proxy wrappers, so it seems like a reasonable compromise. One other change I made at the same time was to rename value->kind for the oneofs in our custom study protos, as 'value' was easily confused with the 'case/value' output that protobuf-es has. With protobuf.js codegen removed, touching a proto file and invoking ./run drops from about 8s to 6s. This closes #2043. * Allow tree-shaking on protobuf types * Display backend error messages in our ts alert() * Make sourcemap generation opt-in for ts-run Considerably slows down build, and not used most of the time.
2023-06-14 14:47:37 +02:00
"anki_io",
"anyhow",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
"camino",
"clap 4.2.1",
Migrate to protobuf-es (#2547) * Fix .no-reduce-motion missing from graphs spinner, and not being honored * Begin migration from protobuf.js -> protobuf-es Motivation: - Protobuf-es has a nicer API: messages are represented as classes, and fields which should exist are not marked as nullable. - As it uses modules, only the proto messages we actually use get included in our bundle output. Protobuf.js put everything in a namespace, which prevented tree-shaking, and made it awkward to access inner messages. - ./run after touching a proto file drops from about 8s to 6s on my machine. The tradeoff is slower decoding/encoding (#2043), but that was mainly a concern for the graphs page, and was unblocked by https://github.com/ankitects/anki/commit/37151213cd9d431f449ba4b3bc4c0329a1d9af78 Approach/notes: - We generate the new protobuf-es interface in addition to existing protobuf.js interface, so we can migrate a module at a time, starting with the graphs module. - rslib:proto now generates RPC methods for TS in addition to the Python interface. The input-arg-unrolling behaviour of the Python generation is not required here, as we declare the input arg as a PlainMessage<T>, which marks it as requiring all fields to be provided. - i64 is represented as bigint in protobuf-es. We were using a patch to protobuf.js to get it to output Javascript numbers instead of long.js types, but now that our supported browser versions support bigint, it's probably worth biting the bullet and migrating to bigint use. Our IDs fit comfortably within MAX_SAFE_INTEGER, but that may not hold for future fields we add. - Oneofs are handled differently in protobuf-es, and are going to need some refactoring. Other notable changes: - Added a --mkdir arg to our build runner, so we can create a dir easily during the build on Windows. - Simplified the preference handling code, by wrapping the preferences in an outer store, instead of a separate store for each individual preference. This means a change to one preference will trigger a redraw of all components that depend on the preference store, but the redrawing is cheap after moving the data processing to Rust, and it makes the code easier to follow. - Drop async(Reactive).ts in favour of more explicit handling with await blocks/updating. - Renamed add_inputs_to_group() -> add_dependency(), and fixed it not adding dependencies to parent groups. Renamed add() -> add_action() for clarity. * Remove a couple of unused proto imports * Migrate card info * Migrate congrats, image occlusion, and tag editor + Fix imports for multi-word proto files. * Migrate change-notetype * Migrate deck options * Bump target to es2020; simplify ts lib list Have used caniuse.com to confirm Chromium 77, iOS 14.5 and the Chrome on Android support the full es2017-es2020 features. * Migrate import-csv * Migrate i18n and fix missing output types in .js * Migrate custom scheduling, and remove protobuf.js To mostly maintain our old API contract, we make use of protobuf-es's ability to convert to JSON, which follows the same format as protobuf.js did. It doesn't cover all case: users who were previously changing the variant of a type will need to update their code, as assigning to a new variant no longer automatically removes the old one, which will cause an error when we try to convert back from JSON. But I suspect the large majority of users are adjusting the current variant rather than creating a new one, and this saves us having to write proxy wrappers, so it seems like a reasonable compromise. One other change I made at the same time was to rename value->kind for the oneofs in our custom study protos, as 'value' was easily confused with the 'case/value' output that protobuf-es has. With protobuf.js codegen removed, touching a proto file and invoking ./run drops from about 8s to 6s. This closes #2043. * Allow tree-shaking on protobuf types * Display backend error messages in our ts alert() * Make sourcemap generation opt-in for ts-run Considerably slows down build, and not used most of the time.
2023-06-14 14:47:37 +02:00
"itertools",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
"junction",
"termcolor",
"which",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
"workspace-hack",
2020-11-24 09:41:03 +01:00
]
[[package]]
name = "rusqlite"
version = "0.29.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "549b9d036d571d42e6e85d1c1425e2ac83491075078ca9a15be021c56b1641f2"
dependencies = [
"bitflags 2.0.2",
"fallible-iterator",
"fallible-streaming-iterator",
"hashlink",
"libsqlite3-sys",
"smallvec",
]
[[package]]
name = "rustc-demangle"
version = "0.1.22"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "d4a36c42d1873f9a77c53bde094f9664d9891bc604a45b4798fd2c389ed12e5b"
2021-02-03 11:29:48 +01:00
[[package]]
name = "rustc-hash"
version = "1.1.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "08d43f7aa6b08d49f382cde6a7982047c3426db949b1424bc4b7ec9ae12c6ce2"
[[package]]
name = "rustix"
version = "0.37.20"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "b96e891d04aa506a6d1f318d2771bcb1c7dfda84e126660ace067c9b474bb2c0"
dependencies = [
"bitflags 1.3.2",
"errno",
"io-lifetimes",
"libc",
"linux-raw-sys",
"windows-sys 0.48.0",
]
[[package]]
name = "rustls"
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
version = "0.20.8"
source = "registry+https://github.com/rust-lang/crates.io-index"
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
checksum = "fff78fc74d175294f4e83b28343315ffcfb114b156f0185e9741cb5570f50e2f"
dependencies = [
"log",
"ring",
"sct",
"webpki",
]
2021-06-18 10:15:41 +02:00
[[package]]
name = "rustls-native-certs"
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
version = "0.6.2"
2021-06-18 10:15:41 +02:00
source = "registry+https://github.com/rust-lang/crates.io-index"
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
checksum = "0167bac7a9f490495f3c33013e7722b53cb087ecbe082fb0c6387c96f634ea50"
2021-06-18 10:15:41 +02:00
dependencies = [
"openssl-probe",
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
"rustls-pemfile",
2021-06-18 10:15:41 +02:00
"schannel",
"security-framework",
]
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
[[package]]
name = "rustls-pemfile"
version = "1.0.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "d194b56d58803a43635bdc398cd17e383d6f71f9182b9a192c127ca42494a59b"
dependencies = [
"base64 0.21.0",
]
2021-10-02 12:42:03 +02:00
[[package]]
name = "rustversion"
version = "1.0.12"
2021-10-02 12:42:03 +02:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "4f3208ce4d8448b3f3e7d168a73f5e0c43a61e32930de3bceeccedb388b6bf06"
2021-10-02 12:42:03 +02:00
[[package]]
name = "ryu"
version = "1.0.13"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "f91339c0467de62360649f8d3e185ca8de4224ff281f66000de5eb2a77a79041"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
[[package]]
name = "safemem"
version = "0.3.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "ef703b7cb59335eae2eb93ceb664c0eb7ea6bf567079d843e09420219668e072"
[[package]]
name = "same-file"
version = "1.0.6"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "93fc1dc3aaa9bfed95e02e6eadabb4baf7e3078b0bd1b4d7b6b0b68378900502"
dependencies = [
"winapi-util",
]
2020-11-03 01:26:27 +01:00
[[package]]
name = "schannel"
version = "0.1.21"
2020-11-03 01:26:27 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "713cfb06c7059f3588fb8044c0fad1d09e3c01d225e25b9220dbfdcf16dbb1b3"
2020-11-03 01:26:27 +01:00
dependencies = [
"windows-sys 0.42.0",
2020-11-03 01:26:27 +01:00
]
[[package]]
name = "scoped-tls"
version = "1.0.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "e1cf6437eb19a8f4a6cc0f7dca544973b0b78843adbfeb3683d1a94a0024a294"
[[package]]
name = "scopeguard"
version = "1.1.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "d29ab0c6d3fc0ee92fe66e2d99f700eab17a8d57d1c1d3b748380fb20baa78cd"
[[package]]
name = "scratch"
version = "1.0.5"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "1792db035ce95be60c3f8853017b3999209281c24e2ba5bc8e59bf97a0c590c1"
[[package]]
name = "sct"
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
version = "0.7.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
checksum = "d53dcdb7c9f8158937a7981b48accfd39a43af418591a5d008c7b22b5e1b7ca4"
dependencies = [
"ring",
"untrusted",
]
2020-11-03 01:26:27 +01:00
[[package]]
name = "security-framework"
version = "2.8.2"
2020-11-03 01:26:27 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "a332be01508d814fed64bf28f798a146d73792121129962fdf335bb3c49a4254"
2020-11-03 01:26:27 +01:00
dependencies = [
"bitflags 1.3.2",
2020-11-03 01:26:27 +01:00
"core-foundation",
"core-foundation-sys",
"libc",
"security-framework-sys",
]
[[package]]
name = "security-framework-sys"
version = "2.8.0"
2020-11-03 01:26:27 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "31c9bb296072e961fcbd8853511dd39c2d8be2deb1e17c6860b1d30732b323b4"
2020-11-03 01:26:27 +01:00
dependencies = [
"core-foundation-sys",
"libc",
]
[[package]]
name = "self_cell"
2021-12-03 11:04:47 +01:00
version = "0.10.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
2021-12-03 11:04:47 +01:00
checksum = "1ef965a420fe14fdac7dd018862966a4c14094f900e1650bbc71ddd7d580c8af"
[[package]]
name = "semver"
version = "1.0.17"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "bebd363326d05ec3e2f532ab7660680f3b02130d780c299bca73469d521bc0ed"
[[package]]
name = "serde"
version = "1.0.164"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "9e8c8cf938e98f769bc164923b06dce91cea1751522f46f8466461af04c9027d"
dependencies = [
"serde_derive",
]
[[package]]
name = "serde-aux"
version = "4.1.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "c599b3fd89a75e0c18d6d2be693ddb12cccaf771db4ff9e39097104808a014c0"
dependencies = [
"chrono",
"serde",
"serde_json",
]
[[package]]
name = "serde_derive"
version = "1.0.164"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "d9735b638ccc51c28bf6914d90a2e9725b377144fc612c49a611fddd1b631d68"
dependencies = [
"proc-macro2",
"quote",
"syn 2.0.12",
]
[[package]]
name = "serde_json"
version = "1.0.95"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "d721eca97ac802aa7777b701877c8004d950fc142651367300d21c1cc0194744"
dependencies = [
"itoa",
"ryu",
"serde",
]
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
[[package]]
name = "serde_path_to_error"
version = "0.1.11"
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "f7f05c1d5476066defcdfacce1f52fc3cae3af1d3089727100c02ae92e5abbe0"
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
dependencies = [
"serde",
]
[[package]]
name = "serde_qs"
version = "0.8.5"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "c7715380eec75f029a4ef7de39a9200e0a63823176b759d055b613f5a87df6a6"
dependencies = [
"percent-encoding",
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
"serde",
"thiserror",
]
[[package]]
name = "serde_repr"
version = "0.1.12"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "bcec881020c684085e55a25f7fd888954d56609ef363479dc5a1305eb0d40cab"
dependencies = [
"proc-macro2",
"quote",
"syn 2.0.12",
]
[[package]]
name = "serde_tuple"
version = "0.5.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "f4f025b91216f15a2a32aa39669329a475733590a015835d1783549a56d09427"
dependencies = [
"serde",
"serde_tuple_macros",
]
[[package]]
name = "serde_tuple_macros"
version = "0.5.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "4076151d1a2b688e25aaf236997933c66e18b870d0369f8b248b8ab2be630d7e"
dependencies = [
"proc-macro2",
"quote",
"syn 1.0.109",
]
[[package]]
name = "serde_urlencoded"
version = "0.7.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "d3491c14715ca2294c4d6a88f15e84739788c1d030eed8c110436aafdaa2f3fd"
dependencies = [
"form_urlencoded",
"itoa",
"ryu",
"serde",
]
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
[[package]]
name = "sha1"
version = "0.10.5"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "f04293dc80c3993519f2d7f6f511707ee7094fe0c6d3406feb330cdb3540eba3"
dependencies = [
"cfg-if",
"cpufeatures",
2022-11-30 03:32:27 +01:00
"digest",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
]
[[package]]
name = "sha2"
version = "0.10.6"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "82e6b795fe2e3b1e845bafcb27aa35405c4d47cdfc92af5fc8d3002f76cebdc0"
dependencies = [
"cfg-if",
"cpufeatures",
2022-11-30 03:32:27 +01:00
"digest",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
]
[[package]]
name = "sharded-slab"
version = "0.1.4"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "900fba806f70c630b0a382d0d825e17a0f19fcd059a2ade1ff237bcddf446b31"
dependencies = [
"lazy_static",
]
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
[[package]]
name = "shared_child"
version = "1.0.0"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "b0d94659ad3c2137fef23ae75b03d5241d633f8acded53d672decfa0e6e0caef"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
dependencies = [
"libc",
"winapi",
]
[[package]]
name = "shlex"
version = "1.1.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "43b2853a4d09f215c24cc5489c992ce46052d359b5109343cbafbf26bc62f8a3"
[[package]]
name = "signal-hook-registry"
version = "1.4.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "d8229b473baa5980ac72ef434c4415e70c4b5e71b423043adb4ba059f89c99a1"
dependencies = [
"libc",
]
2022-11-30 03:32:27 +01:00
[[package]]
name = "simple-file-manifest"
version = "0.11.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "5dd19be0257552dd56d1bb6946f89f193c6e5b9f13cc9327c4bc84a357507c74"
[[package]]
name = "siphasher"
version = "0.3.10"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "7bd3e3206899af3f8b12af284fafc038cc1dc2b41d1b89dd17297221c5d225de"
[[package]]
name = "slab"
version = "0.4.8"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "6528351c9bc8ab22353f9d776db39a20288e8d6c37ef8cfe3317cf875eecfc2d"
2022-09-24 04:39:21 +02:00
dependencies = [
"autocfg",
]
[[package]]
name = "smallvec"
version = "1.10.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "a507befe795404456341dfab10cef66ead4c041f62b8b11bbb92bffe5d0953e0"
[[package]]
name = "snafu"
version = "0.7.4"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "cb0656e7e3ffb70f6c39b3c2a86332bb74aa3c679da781642590f3c1118c5045"
dependencies = [
"backtrace",
"doc-comment",
"snafu-derive",
]
[[package]]
name = "snafu-derive"
version = "0.7.4"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "475b3bbe5245c26f2d8a6f62d67c1f30eb9fffeccee721c45d162c3ebbdf81b2"
dependencies = [
"heck",
"proc-macro2",
"quote",
"syn 1.0.109",
]
V3 parent limits (#1638) * avoid repinning Rust deps by default * add id_tree dependency * Respect intermediate child limits in v3 * Test new behaviour of v3 counts * Rework v3 queue building to respect parent limits * Add missing did field to SQL query * Fix `LimitTreeMap::is_exhausted()` * Rework tree building logic https://github.com/ankitects/anki/pull/1638#discussion_r798328734 * Add timer for build_queues() * `is_exhausted()` -> `limit_reached()` * Move context and limits into `QueueBuilder` This allows for moving more logic into QueueBuilder, so less passing around of arguments. Unfortunately, some tests will require additional work to set up. * Fix stop condition in new_cards_by_position * Fix order gather order of new cards by deck * Add scheduler/queue/builder/burying.rs * Fix bad tree due to unsorted child decks * Fix comment * Fix `cap_new_to_review_rec()` * Add test for new card gathering * Always sort `child_decks()` * Fix deck removal in `cap_new_to_review_rec()` * Fix sibling ordering in new card gathering * Remove limits for deck total count with children * Add random gather order * Remove bad sibling order handling All routines ensure ascending order now. Also do some other minor refactoring. * Remove queue truncating All routines stop now as soon as the root limit is reached. * Move deck fetching into `QueueBuilder::new()` * Rework new card gather and sort options https://github.com/ankitects/anki/pull/1638#issuecomment-1032173013 * Disable new sort order choices ... depending on set gather order. * Use enum instead of numbers * Ensure valid sort order setting * Update new gather and sort order tooltips * Warn about random insertion order with v3 * Revert "Add timer for build_queues()" This reverts commit c9f5fc6ebe87953c17a0c842990b009b5596c69c. * Update rslib/src/storage/card/mod.rs (dae) * minor wording tweaks to the tooltips (dae) + move legacy strings to bottom + consistent capitalization (our leech action still needs fixing, but that will require introducing a new 'suspend card' string as the existing one is used elsewhere as well)
2022-02-10 00:55:43 +01:00
[[package]]
name = "snowflake"
version = "1.3.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "27207bb65232eda1f588cf46db2fee75c0808d557f6b3cf19a75f5d6d7c94df1"
[[package]]
name = "socket2"
version = "0.4.9"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "64a4a911eed85daf18834cfaa86a79b7d266ff93ff5ba14005426219480ed662"
dependencies = [
"libc",
"winapi",
]
[[package]]
name = "spin"
version = "0.5.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "6e63cff320ae2c57904679ba7cb63280a3dc4613885beafb148ee7bf9aa9042d"
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
[[package]]
name = "spin"
version = "0.9.8"
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "6980e8d7511241f8acf4aebddbb1ff938df5eebe98691418c4468d0b72a96a67"
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
[[package]]
name = "string_cache"
version = "0.8.7"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "f91138e76242f575eb1d3b38b4f1362f10d3a43f47d182a5b359af488a02293b"
dependencies = [
"new_debug_unreachable",
2022-09-24 04:39:21 +02:00
"once_cell",
"parking_lot",
"phf_shared 0.10.0",
"precomputed-hash",
"serde",
]
[[package]]
name = "string_cache_codegen"
2022-09-24 04:39:21 +02:00
version = "0.5.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
2022-09-24 04:39:21 +02:00
checksum = "6bb30289b722be4ff74a408c3cc27edeaad656e06cb1fe8fa9231fa59c728988"
dependencies = [
2022-09-24 04:39:21 +02:00
"phf_generator 0.10.0",
"phf_shared 0.10.0",
"proc-macro2",
"quote",
]
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
[[package]]
name = "strsim"
version = "0.10.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "73473c0e59e6d5812c5dfe2a064a6444949f089e20eec9a2e5506596494e4623"
2021-03-07 10:08:03 +01:00
[[package]]
name = "strum"
2022-09-24 04:39:21 +02:00
version = "0.24.1"
2021-03-07 10:08:03 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
2022-09-24 04:39:21 +02:00
checksum = "063e6045c0e62079840579a7e47a355ae92f60eb74daaf156fb1e84ba164e63f"
2021-03-07 10:08:03 +01:00
dependencies = [
"strum_macros",
]
[[package]]
name = "strum_macros"
2022-09-24 04:39:21 +02:00
version = "0.24.3"
2021-03-07 10:08:03 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
2022-09-24 04:39:21 +02:00
checksum = "1e385be0d24f186b4ce2f9982191e7101bb737312ad61c1f2f984f34bcf85d59"
2021-03-07 10:08:03 +01:00
dependencies = [
"heck",
"proc-macro2",
"quote",
2021-11-18 11:54:00 +01:00
"rustversion",
"syn 1.0.109",
2021-03-07 10:08:03 +01:00
]
2022-01-15 05:59:43 +01:00
[[package]]
name = "subtle"
version = "2.4.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "6bdef32e8150c2a081110b42772ffe7d7c9032b606bc226c8260fd97e0976601"
[[package]]
name = "syn"
version = "1.0.109"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "72b64191b275b66ffe2469e8af2c1cfe3bafa67b529ead792a6d0160888b4237"
dependencies = [
"proc-macro2",
"quote",
"unicode-ident",
]
[[package]]
name = "syn"
version = "2.0.12"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "79d9531f94112cfc3e4c8f5f02cb2b58f72c97b7efd85f70203cc6d8efda5927"
dependencies = [
"proc-macro2",
"quote",
2022-09-24 04:39:21 +02:00
"unicode-ident",
]
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
[[package]]
name = "sync_wrapper"
version = "0.1.2"
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "2047c6ded9c721764247e62cd3b03c09ffc529b2ba5b10ec482ae507a4a70160"
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
[[package]]
name = "tar"
version = "0.4.38"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "4b55807c0344e1e6c04d7c965f5289c39a8d94ae23ed5c0b57aabac549f871c6"
dependencies = [
"filetime",
"libc",
"xattr",
]
2022-09-24 04:39:21 +02:00
[[package]]
name = "target-lexicon"
version = "0.12.6"
2022-09-24 04:39:21 +02:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "8ae9980cab1db3fceee2f6c6f643d5d8de2997c58ee8d25fb0cc8a9e9e7348e5"
2022-09-24 04:39:21 +02:00
[[package]]
name = "tempfile"
version = "3.6.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "31c0432476357e58790aaa47a8efb0c5138f137343f3b5f23bd36a27e3b0a6d6"
dependencies = [
"autocfg",
2021-10-02 12:42:03 +02:00
"cfg-if",
2022-01-15 05:59:43 +01:00
"fastrand",
"redox_syscall 0.3.5",
"rustix",
"windows-sys 0.48.0",
]
[[package]]
name = "tendril"
2022-09-24 04:39:21 +02:00
version = "0.4.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
2022-09-24 04:39:21 +02:00
checksum = "d24a120c5fc464a3458240ee02c299ebcb9d67b5249c8848b09d639dca8d7bb0"
dependencies = [
"futf",
"mac",
"utf-8",
]
[[package]]
name = "termcolor"
version = "1.2.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "be55cf8942feac5c765c2c993422806843c9a9a45d4d5c407ad6dd2ea95eb9b6"
dependencies = [
"winapi-util",
]
[[package]]
name = "terminal_size"
version = "0.2.6"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "8e6bf6f19e9f8ed8d4048dc22981458ebcf406d67e94cd422e5ecd73d63b3237"
dependencies = [
"rustix",
"windows-sys 0.48.0",
]
[[package]]
name = "textwrap"
version = "0.16.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "222a222a5bfe1bba4a77b45ec488a741b3cb8872e5e499451fd7d0129c9c7c3d"
[[package]]
name = "thiserror"
version = "1.0.40"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "978c9a314bd8dc99be594bc3c175faaa9794be04a5a5e153caba6915336cebac"
dependencies = [
"thiserror-impl",
]
[[package]]
name = "thiserror-impl"
version = "1.0.40"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "f9456a42c5b0d803c8cd86e73dd7cc9edd429499f37a3550d286d5e86720569f"
dependencies = [
"proc-macro2",
"quote",
"syn 2.0.12",
]
[[package]]
name = "thread_local"
version = "1.1.7"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "3fdd6f064ccff2d6567adcb3873ca630700f00b5ad3f060c25b5dcfd9a4ce152"
dependencies = [
"cfg-if",
2021-02-03 11:29:48 +01:00
"once_cell",
]
[[package]]
name = "time"
version = "0.3.20"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "cd0cbfecb4d19b5ea75bb31ad904eb5b9fa13f21079c3b92017ebdf4999a5890"
dependencies = [
"itoa",
2022-11-30 03:32:27 +01:00
"serde",
"time-core",
"time-macros",
]
2022-11-30 03:32:27 +01:00
[[package]]
name = "time-core"
version = "0.1.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "2e153e1f1acaef8acc537e68b44906d2db6436e2b35ac2c6b42640fff91f00fd"
[[package]]
name = "time-macros"
version = "0.2.8"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "fd80a657e71da814b8e5d60d3374fc6d35045062245d80224748ae522dd76f36"
2022-11-30 03:32:27 +01:00
dependencies = [
"time-core",
]
[[package]]
name = "tinystr"
version = "0.7.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "7ac3f5b6856e931e15e07b478e98c8045239829a65f9156d4fa7e7788197a5ef"
dependencies = [
"displaydoc",
]
[[package]]
name = "tinytemplate"
version = "1.2.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "be4d6b5f19ff7664e8c98d03e2139cb510db9b0a60b55f8e8709b689d939b6bc"
dependencies = [
"serde",
"serde_json",
]
[[package]]
name = "tinyvec"
2022-09-24 04:39:21 +02:00
version = "1.6.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
2022-09-24 04:39:21 +02:00
checksum = "87cc5ceb3875bb20c2890005a4e226a4651264a5c75edb2421b52861a0a0cb50"
2020-11-12 08:35:14 +01:00
dependencies = [
"tinyvec_macros",
]
[[package]]
name = "tinyvec_macros"
version = "0.1.1"
2020-11-12 08:35:14 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "1f3ccbac311fea05f86f61904b462b55fb3df8837a366dfc601a0161d0532f20"
[[package]]
name = "tokio"
version = "1.27.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "d0de47a4eecbe11f498978a9b29d792f0d2692d1dd003650c24c76510e3bc001"
dependencies = [
2022-09-24 04:39:21 +02:00
"autocfg",
"bytes",
"libc",
"mio",
"num_cpus",
2022-09-24 04:39:21 +02:00
"parking_lot",
"pin-project-lite",
"signal-hook-registry",
"socket2",
"tokio-macros",
"windows-sys 0.45.0",
]
[[package]]
name = "tokio-macros"
version = "2.0.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "61a573bdc87985e9d6ddeed1b3d864e8a302c847e40d647746df2f1de209d1ce"
dependencies = [
"proc-macro2",
"quote",
"syn 2.0.12",
]
[[package]]
name = "tokio-native-tls"
version = "0.3.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "bbae76ab933c85776efabc971569dd6119c580d8f5d448769dec1764bf796ef2"
dependencies = [
"native-tls",
"tokio",
]
[[package]]
name = "tokio-rustls"
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
version = "0.23.4"
source = "registry+https://github.com/rust-lang/crates.io-index"
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
checksum = "c43ee83903113e03984cb9e5cebe6c04a5116269e900e3ddba8f068a62adda59"
dependencies = [
"rustls",
"tokio",
"webpki",
]
[[package]]
name = "tokio-socks"
version = "0.5.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "51165dfa029d2a65969413a6cc96f354b86b464498702f174a4efa13608fd8c0"
dependencies = [
"either",
"futures-util",
"thiserror",
"tokio",
]
[[package]]
name = "tokio-stream"
version = "0.1.12"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "8fb52b74f05dbf495a8fba459fdc331812b96aa086d9eb78101fa0d4569c3313"
dependencies = [
"futures-core",
"pin-project-lite",
"tokio",
]
[[package]]
name = "tokio-tungstenite"
version = "0.18.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "54319c93411147bced34cb5609a80e0a8e44c5999c93903a81cd866630ec0bfd"
dependencies = [
"futures-util",
"log",
"tokio",
"tungstenite",
]
[[package]]
name = "tokio-util"
version = "0.7.7"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "5427d89453009325de0d8f342c9490009f76e999cb7672d77e46267448f7e6b2"
dependencies = [
"bytes",
"futures-core",
"futures-sink",
"pin-project-lite",
"tokio",
2022-09-24 04:39:21 +02:00
"tracing",
]
[[package]]
name = "toml"
version = "0.5.11"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "f4f7f0dd8d50a853a531c426359045b1998f04219d88799810762cd4ad314234"
dependencies = [
"serde",
]
[[package]]
name = "toml_datetime"
version = "0.6.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "3ab8ed2edee10b50132aed5f331333428b011c99402b5a534154ed15746f9622"
[[package]]
name = "toml_edit"
version = "0.19.8"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "239410c8609e8125456927e6707163a3b1fdb40561e4b803bc041f466ccfdc13"
dependencies = [
"indexmap",
"toml_datetime",
"winnow",
]
[[package]]
name = "topological-sort"
version = "0.2.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "ea68304e134ecd095ac6c3574494fc62b909f416c4fca77e440530221e549d3d"
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
[[package]]
name = "tower"
version = "0.4.13"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "b8fa9be0de6cf49e536ce1851f987bd21a43b771b09473c3549a6c853db37c1c"
dependencies = [
"futures-core",
"futures-util",
"pin-project",
"pin-project-lite",
"tokio",
"tower-layer",
"tower-service",
"tracing",
]
[[package]]
name = "tower-http"
version = "0.4.0"
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "5d1d42a9b3f3ec46ba828e8d376aec14592ea199f70a06a548587ecd1c4ab658"
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
dependencies = [
"bitflags 1.3.2",
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
"bytes",
"futures-core",
"futures-util",
"http",
"http-body",
"http-range-header",
"pin-project-lite",
"tower-layer",
"tower-service",
"tracing",
]
[[package]]
name = "tower-layer"
version = "0.3.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "c20c8dbed6283a09604c3e69b4b7eeb54e298b8a600d4d5ecb5ad39de609f1d0"
[[package]]
name = "tower-service"
2022-09-24 04:39:21 +02:00
version = "0.3.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
2022-09-24 04:39:21 +02:00
checksum = "b6bc1c9ce2b5135ac7f93c72918fc37feb872bdc6a5533a8b85eb4b86bfdae52"
[[package]]
name = "tracing"
version = "0.1.37"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "8ce8c33a8d48bd45d624a6e523445fd21ec13d3653cd51f681abf67418f54eb8"
dependencies = [
2021-10-02 12:42:03 +02:00
"cfg-if",
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
"log",
"pin-project-lite",
"tracing-attributes",
"tracing-core",
]
[[package]]
name = "tracing-appender"
version = "0.2.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "09d48f71a791638519505cefafe162606f706c25592e4bde4d97600c0195312e"
dependencies = [
"crossbeam-channel",
"time",
"tracing-subscriber",
]
[[package]]
name = "tracing-attributes"
version = "0.1.23"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "4017f8f45139870ca7e672686113917c71c7a6e02d4924eda67186083c03081a"
dependencies = [
"proc-macro2",
"quote",
"syn 1.0.109",
]
[[package]]
name = "tracing-core"
version = "0.1.30"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "24eb03ba0eab1fd845050058ce5e616558e8f8d8fca633e6b163fe25c797213a"
dependencies = [
2022-09-24 04:39:21 +02:00
"once_cell",
"valuable",
]
[[package]]
name = "tracing-log"
version = "0.1.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "78ddad33d2d10b1ed7eb9d1f518a5674713876e97e5bb9b7345a7984fbb4f922"
dependencies = [
"lazy_static",
"log",
"tracing-core",
]
[[package]]
name = "tracing-subscriber"
version = "0.3.16"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "a6176eae26dd70d0c919749377897b54a9276bd7061339665dd68777926b5a70"
dependencies = [
"matchers",
"nu-ansi-term",
"once_cell",
"regex",
"sharded-slab",
"smallvec",
"thread_local",
"tracing",
"tracing-core",
"tracing-log",
]
[[package]]
name = "try-lock"
version = "0.2.4"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "3528ecfd12c466c6f163363caf2d02a71161dd5e1cc6ae7b34207ea2d42d81ed"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
[[package]]
name = "tugger-common"
version = "0.10.0"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "f90d950380afdb1a6bbe74f29485a04e821869dfad11f5929ff1c5b1dac09d02"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
dependencies = [
"anyhow",
"fs2",
"glob",
"hex",
2022-11-30 03:32:27 +01:00
"log",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
"once_cell",
"reqwest",
2022-11-30 03:32:27 +01:00
"sha2",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
"tempfile",
"url",
2022-11-30 03:32:27 +01:00
"zip",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
]
[[package]]
name = "tugger-windows"
version = "0.10.0"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "e9f181ac4fc7f8facfd418824d13045cd068ee73de44319a6116868c22789782"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
dependencies = [
"anyhow",
"duct",
"find-winsdk",
"glob",
"once_cell",
"semver",
"tugger-common",
"winapi",
]
[[package]]
name = "tugger-windows-codesign"
version = "0.10.0"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "ed3f09f8bdace495373cec3fc607bc39f00720a984ba82e310cc9382462fd364"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
dependencies = [
"anyhow",
"duct",
2022-11-30 03:32:27 +01:00
"log",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
"p12",
"rcgen",
"time",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
"tugger-common",
"tugger-windows",
"yasna",
]
[[package]]
name = "tungstenite"
version = "0.18.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "30ee6ab729cd4cf0fd55218530c4522ed30b7b6081752839b68fcec8d0960788"
dependencies = [
"base64 0.13.1",
"byteorder",
"bytes",
"http",
"httparse",
"log",
"rand 0.8.5",
"sha1",
"thiserror",
"url",
"utf-8",
]
[[package]]
name = "type-map"
2021-02-03 11:29:48 +01:00
version = "0.4.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
2021-02-03 11:29:48 +01:00
checksum = "b6d3364c5e96cb2ad1603037ab253ddd34d7fb72a58bdddf4b7350760fc69a46"
dependencies = [
2021-02-03 11:29:48 +01:00
"rustc-hash",
]
[[package]]
name = "typenum"
version = "1.16.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "497961ef93d974e23eb6f433eb5fe1b7930b659f06d12dec6fc44a8f554c0bba"
[[package]]
name = "ucd-trie"
version = "0.1.5"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "9e79c4d996edb816c91e4308506774452e55e95c3c9de07b6729e17e15a5ef81"
[[package]]
name = "unic-char-property"
version = "0.9.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "a8c57a407d9b6fa02b4795eb81c5b6652060a15a7903ea981f3d723e6c0be221"
dependencies = [
"unic-char-range",
]
[[package]]
name = "unic-char-range"
version = "0.9.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "0398022d5f700414f6b899e10b8348231abf9173fa93144cbc1a43b9793c1fbc"
[[package]]
name = "unic-common"
version = "0.9.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "80d7ff825a6a654ee85a63e80f92f054f904f21e7d12da4e22f9834a4aaa35bc"
[[package]]
name = "unic-langid"
version = "0.9.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "398f9ad7239db44fd0f80fe068d12ff22d78354080332a5077dc6f52f14dcf2f"
dependencies = [
"unic-langid-impl",
"unic-langid-macros",
]
[[package]]
name = "unic-langid-impl"
version = "0.9.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "e35bfd2f2b8796545b55d7d3fd3e89a0613f68a0d1c8bc28cb7ff96b411a35ff"
dependencies = [
"tinystr",
]
[[package]]
name = "unic-langid-macros"
version = "0.9.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "055e618bf694161ffff0466d95cef3e1a5edc59f6ba1888e97801f2b4ebdc4fe"
dependencies = [
"proc-macro-hack",
"tinystr",
"unic-langid-impl",
"unic-langid-macros-impl",
]
[[package]]
name = "unic-langid-macros-impl"
version = "0.9.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "1f5cdec05b907f4e2f6843f4354f4ce6a5bebe1a56df320a49134944477ce4d8"
dependencies = [
"proc-macro-hack",
"quote",
"syn 1.0.109",
"unic-langid-impl",
]
[[package]]
name = "unic-ucd-category"
version = "0.9.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "1b8d4591f5fcfe1bd4453baaf803c40e1b1e69ff8455c47620440b46efef91c0"
dependencies = [
"matches",
"unic-char-property",
"unic-char-range",
"unic-ucd-version",
]
[[package]]
name = "unic-ucd-version"
version = "0.9.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "96bd2f2237fe450fcd0a1d2f5f4e91711124f7857ba2e964247776ebeeb7b0c4"
dependencies = [
"unic-common",
]
[[package]]
name = "unicase"
version = "2.6.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "50f37be617794602aabbeee0be4f259dc1778fabe05e2d67ee8f79326d5cb4f6"
dependencies = [
"version_check",
]
[[package]]
name = "unicode-bidi"
version = "0.3.13"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "92888ba5573ff080736b3648696b70cafad7d250551175acbaa4e0385b3e1460"
2022-09-24 04:39:21 +02:00
[[package]]
name = "unicode-ident"
version = "1.0.8"
2022-09-24 04:39:21 +02:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "e5464a87b239f13a63a501f2701565754bae92d243d4bb7eb12f6d57d2269bf4"
[[package]]
name = "unicode-normalization"
2022-09-24 04:39:21 +02:00
version = "0.1.22"
source = "registry+https://github.com/rust-lang/crates.io-index"
2022-09-24 04:39:21 +02:00
checksum = "5c5713f0fc4b5db668a2ac63cdb7bb4469d8c9fed047b1d0292cc7b0ce2ba921"
dependencies = [
"tinyvec",
]
[[package]]
name = "unicode-segmentation"
version = "1.10.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "1dd624098567895118886609431a7c3b8f516e41d30e0643f03d94592a147e36"
[[package]]
name = "unicode-width"
2022-09-24 04:39:21 +02:00
version = "0.1.10"
source = "registry+https://github.com/rust-lang/crates.io-index"
2022-09-24 04:39:21 +02:00
checksum = "c0edd1e5b14653f783770bce4a4dabb4a5108a5370a5f5d8cfe8710c361f6c8b"
[[package]]
name = "unindent"
version = "0.1.11"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "e1766d682d402817b5ac4490b3c3002d91dfa0d22812f341609f97b08757359c"
[[package]]
name = "untrusted"
version = "0.7.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "a156c684c91ea7d62626509bce3cb4e1d9ed5c4d978f7b4352658f96a4c26b4a"
[[package]]
name = "url"
2022-09-24 04:39:21 +02:00
version = "2.3.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
2022-09-24 04:39:21 +02:00
checksum = "0d68c799ae75762b8c3fe375feb6600ef5602c883c5d21eb51c09f22b83c4643"
dependencies = [
"form_urlencoded",
"idna",
"percent-encoding",
"serde",
]
[[package]]
name = "utf-8"
version = "0.7.6"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "09cc8ee72d2a9becf2f2febe0205bbed8fc6615b7cb429ad062dc7b7ddd036a9"
[[package]]
name = "utf8parse"
version = "0.2.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "711b9620af191e0cdc7468a8d14e709c3dcdb115b36f838e601583af800a370a"
[[package]]
name = "utime"
version = "0.3.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "91baa0c65eabd12fcbdac8cc35ff16159cab95cae96d0222d6d0271db6193cef"
dependencies = [
"libc",
"winapi",
]
[[package]]
name = "valuable"
version = "0.1.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "830b7e5d4d90034032940e4ace0d9a9a057e7a45cd94e6c007832e39edb82f6d"
[[package]]
name = "vcpkg"
2021-06-25 07:35:25 +02:00
version = "0.2.15"
source = "registry+https://github.com/rust-lang/crates.io-index"
2021-06-25 07:35:25 +02:00
checksum = "accd4ea62f7bb7a82fe23066fb0957d48ef677f6eeb8215f372f52e48bb32426"
[[package]]
name = "version_check"
2022-01-15 05:59:43 +01:00
version = "0.9.4"
source = "registry+https://github.com/rust-lang/crates.io-index"
2022-01-15 05:59:43 +01:00
checksum = "49874b5167b65d7193b8aba1567f5c7d93d001cafc34600cee003eda787e483f"
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
[[package]]
name = "waker-fn"
version = "1.1.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "9d5b2c62b4012a3e1eca5a7e077d13b3bf498c4073e33ccd58626607748ceeca"
[[package]]
name = "walkdir"
version = "2.3.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "36df944cda56c7d8d8b7496af378e6b16de9284591917d307c9b4d313c44e698"
dependencies = [
"same-file",
"winapi-util",
]
[[package]]
name = "want"
version = "0.3.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "1ce8a968cb1cd110d136ff8b819a556d6fb6d919363c61534f6860c7eb172ba0"
dependencies = [
"log",
"try-lock",
]
[[package]]
name = "warp"
version = "0.3.4"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "27e1a710288f0f91a98dd8a74f05b76a10768db245ce183edf64dc1afdc3016c"
dependencies = [
"bytes",
"futures-channel",
"futures-util",
"headers",
"http",
"hyper",
"log",
"mime",
"mime_guess",
"percent-encoding",
"pin-project",
"rustls-pemfile",
"scoped-tls",
"serde",
"serde_json",
"serde_urlencoded",
"tokio",
"tokio-stream",
"tokio-tungstenite",
"tokio-util",
"tower-service",
"tracing",
]
[[package]]
name = "wasi"
version = "0.9.0+wasi-snapshot-preview1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "cccddf32554fecc6acb585f82a32a72e28b48f8c4c1883ddfeeeaa96f7d8e519"
[[package]]
name = "wasi"
version = "0.11.0+wasi-snapshot-preview1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "9c8d87e72b64a3b4db28d11ce29237c246188f4f51057d65a7eab63b7987e423"
[[package]]
name = "wasm-bindgen"
version = "0.2.84"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "31f8dcbc21f30d9b8f2ea926ecb58f6b91192c17e9d33594b3df58b2007ca53b"
dependencies = [
2021-10-02 12:42:03 +02:00
"cfg-if",
"wasm-bindgen-macro",
]
[[package]]
name = "wasm-bindgen-backend"
version = "0.2.84"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "95ce90fd5bcc06af55a641a86428ee4229e44e07033963a2290a8e241607ccb9"
dependencies = [
"bumpalo",
"log",
2022-09-24 04:39:21 +02:00
"once_cell",
"proc-macro2",
"quote",
"syn 1.0.109",
"wasm-bindgen-shared",
]
[[package]]
name = "wasm-bindgen-futures"
version = "0.4.34"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "f219e0d211ba40266969f6dbdd90636da12f75bee4fc9d6c23d1260dadb51454"
dependencies = [
2021-10-02 12:42:03 +02:00
"cfg-if",
"js-sys",
"wasm-bindgen",
"web-sys",
]
[[package]]
name = "wasm-bindgen-macro"
version = "0.2.84"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "4c21f77c0bedc37fd5dc21f897894a5ca01e7bb159884559461862ae90c0b4c5"
dependencies = [
"quote",
"wasm-bindgen-macro-support",
]
[[package]]
name = "wasm-bindgen-macro-support"
version = "0.2.84"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "2aff81306fcac3c7515ad4e177f521b5c9a15f2b08f4e32d823066102f35a5f6"
dependencies = [
"proc-macro2",
"quote",
"syn 1.0.109",
"wasm-bindgen-backend",
"wasm-bindgen-shared",
]
[[package]]
name = "wasm-bindgen-shared"
version = "0.2.84"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "0046fef7e28c3804e5e38bfa31ea2a0f73905319b677e57ebe37e49358989b5d"
[[package]]
name = "wasm-streams"
version = "0.2.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "6bbae3363c08332cadccd13b67db371814cd214c2524020932f0804b8cf7c078"
dependencies = [
"futures-util",
"js-sys",
"wasm-bindgen",
"wasm-bindgen-futures",
"web-sys",
]
[[package]]
name = "web-sys"
version = "0.3.61"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "e33b99f4b23ba3eec1a53ac264e35a755f00e966e0065077d6027c0f575b0b97"
dependencies = [
"js-sys",
"wasm-bindgen",
]
[[package]]
name = "webpki"
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
version = "0.22.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
checksum = "f095d78192e208183081cc07bc5515ef55216397af48b873e5edcd72637fa1bd"
dependencies = [
"ring",
"untrusted",
]
[[package]]
name = "webpki-roots"
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
version = "0.22.6"
source = "registry+https://github.com/rust-lang/crates.io-index"
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
checksum = "b6c71e40d7d2c34a5106301fb632274ca37242cd0c9d3e64dbece371a40a2d87"
dependencies = [
"webpki",
]
[[package]]
name = "which"
version = "4.4.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "2441c784c52b289a054b7201fc93253e288f094e2f4be9058343127c4226a269"
dependencies = [
"either",
"libc",
2022-09-24 04:39:21 +02:00
"once_cell",
]
[[package]]
name = "winapi"
version = "0.3.9"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "5c839a674fcd7a98952e593242ea400abe93992746761e38641405d28b00f419"
dependencies = [
"winapi-i686-pc-windows-gnu",
"winapi-x86_64-pc-windows-gnu",
]
[[package]]
name = "winapi-i686-pc-windows-gnu"
version = "0.4.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "ac3b87c63620426dd9b991e5ce0329eff545bccbbb34f3be09ff6fb6ab51b7b6"
[[package]]
name = "winapi-util"
version = "0.1.5"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "70ec6ce85bb158151cae5e5c87f95a8e97d2c0c4b001223f33a334e3ce5de178"
dependencies = [
"winapi",
]
[[package]]
name = "winapi-x86_64-pc-windows-gnu"
version = "0.4.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "712e227841d057c1ee1cd2fb22fa7e5a5461ae8e48fa2ca79ec42cfc1931183f"
[[package]]
name = "windows"
version = "0.48.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "e686886bc078bc1b0b600cac0147aadb815089b6e4da64016cbd754b6342700f"
dependencies = [
"windows-targets 0.48.0",
]
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
[[package]]
name = "windows-sys"
version = "0.42.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "5a3e1820f08b8513f676f7ab6c1f99ff312fb97b553d30ff4dd86f9f15728aa7"
dependencies = [
"windows_aarch64_gnullvm 0.42.2",
"windows_aarch64_msvc 0.42.2",
"windows_i686_gnu 0.42.2",
"windows_i686_msvc 0.42.2",
"windows_x86_64_gnu 0.42.2",
"windows_x86_64_gnullvm 0.42.2",
"windows_x86_64_msvc 0.42.2",
]
[[package]]
name = "windows-sys"
version = "0.45.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "75283be5efb2831d37ea142365f009c02ec203cd29a3ebecbc093d52315b66d0"
dependencies = [
"windows-targets 0.42.2",
]
[[package]]
name = "windows-sys"
version = "0.48.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "677d2418bec65e3338edb076e806bc1ec15693c5d0104683f2efe857f61056a9"
dependencies = [
"windows-targets 0.48.0",
]
[[package]]
name = "windows-targets"
version = "0.42.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "8e5180c00cd44c9b1c88adb3693291f1cd93605ded80c250a75d472756b4d071"
dependencies = [
"windows_aarch64_gnullvm 0.42.2",
"windows_aarch64_msvc 0.42.2",
"windows_i686_gnu 0.42.2",
"windows_i686_msvc 0.42.2",
"windows_x86_64_gnu 0.42.2",
"windows_x86_64_gnullvm 0.42.2",
"windows_x86_64_msvc 0.42.2",
]
[[package]]
name = "windows-targets"
version = "0.48.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "7b1eb6f0cd7c80c79759c929114ef071b87354ce476d9d94271031c0497adfd5"
dependencies = [
"windows_aarch64_gnullvm 0.48.0",
"windows_aarch64_msvc 0.48.0",
"windows_i686_gnu 0.48.0",
"windows_i686_msvc 0.48.0",
"windows_x86_64_gnu 0.48.0",
"windows_x86_64_gnullvm 0.48.0",
"windows_x86_64_msvc 0.48.0",
]
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
[[package]]
name = "windows_aarch64_gnullvm"
version = "0.42.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "597a5118570b68bc08d8d59125332c54f1ba9d9adeedeef5b99b02ba2b0698f8"
[[package]]
name = "windows_aarch64_gnullvm"
version = "0.48.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "91ae572e1b79dba883e0d315474df7305d12f569b400fcf90581b06062f7e1bc"
[[package]]
name = "windows_aarch64_msvc"
version = "0.42.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "e08e8864a60f06ef0d0ff4ba04124db8b0fb3be5776a5cd47641e942e58c4d43"
[[package]]
name = "windows_aarch64_msvc"
version = "0.48.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "b2ef27e0d7bdfcfc7b868b317c1d32c641a6fe4629c171b8928c7b08d98d7cf3"
[[package]]
name = "windows_i686_gnu"
version = "0.42.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "c61d927d8da41da96a81f029489353e68739737d3beca43145c8afec9a31a84f"
[[package]]
name = "windows_i686_gnu"
version = "0.48.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "622a1962a7db830d6fd0a69683c80a18fda201879f0f447f065a3b7467daa241"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
[[package]]
name = "windows_i686_msvc"
version = "0.42.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "44d840b6ec649f480a41c8d80f9c65108b92d89345dd94027bfe06ac444d1060"
[[package]]
name = "windows_i686_msvc"
version = "0.48.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "4542c6e364ce21bf45d69fdd2a8e455fa38d316158cfd43b3ac1c5b1b19f8e00"
[[package]]
name = "windows_x86_64_gnu"
version = "0.42.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "8de912b8b8feb55c064867cf047dda097f92d51efad5b491dfb98f6bbb70cb36"
[[package]]
name = "windows_x86_64_gnu"
version = "0.48.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "ca2b8a661f7628cbd23440e50b05d705db3686f894fc9580820623656af974b1"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
[[package]]
name = "windows_x86_64_gnullvm"
version = "0.42.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "26d41b46a36d453748aedef1486d5c7a85db22e56aff34643984ea85514e94a3"
[[package]]
name = "windows_x86_64_gnullvm"
version = "0.48.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "7896dbc1f41e08872e9d5e8f8baa8fdd2677f29468c4e156210174edc7f7b953"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
[[package]]
name = "windows_x86_64_msvc"
version = "0.42.2"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "9aec5da331524158c6d1a4ac0ab1541149c0b9505fde06423b02f5ef0106b9f0"
[[package]]
name = "windows_x86_64_msvc"
version = "0.48.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "1a515f5799fe4961cb532f983ce2b23082366b898e52ffbce459c86f67c8378a"
[[package]]
name = "winnow"
version = "0.4.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "ae8970b36c66498d8ff1d66685dc86b91b29db0c7739899012f63a63814b4b28"
dependencies = [
"memchr",
]
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
[[package]]
name = "winreg"
version = "0.5.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "a27a759395c1195c4cc5cda607ef6f8f6498f64e78f7900f5de0a127a424704a"
dependencies = [
"serde",
"winapi",
]
[[package]]
name = "winreg"
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
version = "0.10.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
checksum = "80d0f4e272c85def139476380b12f9ac60926689dd2e01d4923222f40580869d"
dependencies = [
"winapi",
]
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
[[package]]
name = "wiremock"
version = "0.5.17"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "12316b50eb725e22b2f6b9c4cbede5b7b89984274d113a7440c86e5c3fc6f99b"
dependencies = [
"assert-json-diff",
"async-trait",
"base64 0.13.1",
"deadpool",
"futures",
"futures-timer",
"http-types",
"hyper",
"log",
"once_cell",
"regex",
"serde",
"serde_json",
"tokio",
]
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
[[package]]
name = "workspace-hack"
version = "0.1.0"
dependencies = [
"bytes",
"cc",
2023-05-29 09:54:00 +02:00
"clap 4.2.1",
"clap_builder",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
"crossbeam-utils",
"either",
"flate2",
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
"futures-channel",
"futures-io",
"futures-util",
"getrandom 0.2.8",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
"hashbrown",
2022-11-30 03:32:27 +01:00
"hmac",
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
"hyper",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
"indexmap",
"itertools",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
"log",
"num-traits",
"phf_shared 0.11.1",
"prost",
"rand 0.8.5",
"rand_core 0.6.4",
"regex",
"regex-syntax",
"scopeguard",
"serde",
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
"serde_json",
2022-11-30 03:32:27 +01:00
"sha2",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
"snafu",
"syn 1.0.109",
"syn 2.0.12",
"time",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
"tokio",
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
"tokio-util",
"tracing",
"tracing-core",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
"url",
2022-11-30 03:32:27 +01:00
"zip",
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
"zstd 0.11.2+zstd.1.5.2",
"zstd 0.12.3+zstd.1.5.2",
Rework syncing code, and replace local sync server (#2329) This PR replaces the existing Python-driven sync server with a new one in Rust. The new server supports both collection and media syncing, and is compatible with both the new protocol mentioned below, and older clients. A setting has been added to the preferences screen to point Anki to a local server, and a similar setting is likely to come to AnkiMobile soon. Documentation is available here: <https://docs.ankiweb.net/sync-server.html> In addition to the new server and refactoring, this PR also makes changes to the sync protocol. The existing sync protocol places payloads and metadata inside a multipart POST body, which causes a few headaches: - Legacy clients build the request in a non-deterministic order, meaning the entire request needs to be scanned to extract the metadata. - Reqwest's multipart API directly writes the multipart body, without exposing the resulting stream to us, making it harder to track the progress of the transfer. We've been relying on a patched version of reqwest for timeouts, which is a pain to keep up to date. To address these issues, the metadata is now sent in a HTTP header, with the data payload sent directly in the body. Instead of the slower gzip, we now use zstd. The old timeout handling code has been replaced with a new implementation that wraps the request and response body streams to track progress, allowing us to drop the git dependencies for reqwest, hyper-timeout and tokio-io-timeout. The main other change to the protocol is that one-way syncs no longer need to downgrade the collection to schema 11 prior to sending.
2023-01-18 03:43:46 +01:00
"zstd-safe 5.0.2+zstd.1.5.2",
"zstd-safe 6.0.4+zstd.1.5.4",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
"zstd-sys",
]
[[package]]
name = "xattr"
version = "0.2.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "6d1526bbe5aaeb5eb06885f4d987bcdfa5e23187055de9b83fe00156a821fabc"
dependencies = [
"libc",
]
[[package]]
name = "xml5ever"
version = "0.17.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "4034e1d05af98b51ad7214527730626f019682d797ba38b51689212118d8e650"
dependencies = [
"log",
"mac",
"markup5ever",
]
[[package]]
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
name = "xz2"
version = "0.1.7"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "388c44dc09d76f1536602ead6d325eb532f5c122f17782bd57fb47baeeb767e2"
dependencies = [
"lzma-sys",
]
[[package]]
name = "yasna"
version = "0.5.1"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "aed2e7a52e3744ab4d0c05c20aa065258e84c49fd4226f5191b2ed29712710b4"
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
dependencies = [
"time",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
]
[[package]]
name = "zip"
version = "0.6.4"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "0445d0fbc924bb93539b4316c11afb121ea39296f99a3c4c9edad09e3658cdef"
dependencies = [
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
"aes",
"byteorder",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
"bzip2",
"constant_time_eq 0.1.5",
"crc32fast",
2022-09-24 04:39:21 +02:00
"crossbeam-utils",
"flate2",
2022-11-30 03:32:27 +01:00
"hmac",
Move away from Bazel (#2202) (for upgrading users, please see the notes at the bottom) Bazel brought a lot of nice things to the table, such as rebuilds based on content changes instead of modification times, caching of build products, detection of incorrect build rules via a sandbox, and so on. Rewriting the build in Bazel was also an opportunity to improve on the Makefile-based build we had prior, which was pretty poor: most dependencies were external or not pinned, and the build graph was poorly defined and mostly serialized. It was not uncommon for fresh checkouts to fail due to floating dependencies, or for things to break when trying to switch to an older commit. For day-to-day development, I think Bazel served us reasonably well - we could generally switch between branches while being confident that builds would be correct and reasonably fast, and not require full rebuilds (except on Windows, where the lack of a sandbox and the TS rules would cause build breakages when TS files were renamed/removed). Bazel achieves that reliability by defining rules for each programming language that define how source files should be turned into outputs. For the rules to work with Bazel's sandboxing approach, they often have to reimplement or partially bypass the standard tools that each programming language provides. The Rust rules call Rust's compiler directly for example, instead of using Cargo, and the Python rules extract each PyPi package into a separate folder that gets added to sys.path. These separate language rules allow proper declaration of inputs and outputs, and offer some advantages such as caching of build products and fine-grained dependency installation. But they also bring some downsides: - The rules don't always support use-cases/platforms that the standard language tools do, meaning they need to be patched to be used. I've had to contribute a number of patches to the Rust, Python and JS rules to unblock various issues. - The dependencies we use with each language sometimes make assumptions that do not hold in Bazel, meaning they either need to be pinned or patched, or the language rules need to be adjusted to accommodate them. I was hopeful that after the initial setup work, things would be relatively smooth-sailing. Unfortunately, that has not proved to be the case. Things frequently broke when dependencies or the language rules were updated, and I began to get frustrated at the amount of Anki development time I was instead spending on build system upkeep. It's now about 2 years since switching to Bazel, and I think it's time to cut losses, and switch to something else that's a better fit. The new build system is based on a small build tool called Ninja, and some custom Rust code in build/. This means that to build Anki, Bazel is no longer required, but Ninja and Rust need to be installed on your system. Python and Node toolchains are automatically downloaded like in Bazel. This new build system should result in faster builds in some cases: - Because we're using cargo to build now, Rust builds are able to take advantage of pipelining and incremental debug builds, which we didn't have with Bazel. It's also easier to override the default linker on Linux/macOS, which can further improve speeds. - External Rust crates are now built with opt=1, which improves performance of debug builds. - Esbuild is now used to transpile TypeScript, instead of invoking the TypeScript compiler. This results in faster builds, by deferring typechecking to test/check time, and by allowing more work to happen in parallel. As an example of the differences, when testing with the mold linker on Linux, adding a new message to tags.proto (which triggers a recompile of the bulk of the Rust and TypeScript code) results in a compile that goes from about 22s on Bazel to about 7s in the new system. With the standard linker, it's about 9s. Some other changes of note: - Our Rust workspace now uses cargo-hakari to ensure all packages agree on available features, preventing unnecessary rebuilds. - pylib/anki is now a PEP420 implicit namespace, avoiding the need to merge source files and generated files into a single folder for running. By telling VSCode about the extra search path, code completion now works with generated files without needing to symlink them into the source folder. - qt/aqt can't use PEP420 as it's difficult to get rid of aqt/__init__.py. Instead, the generated files are now placed in a separate _aqt package that's added to the path. - ts/lib is now exposed as @tslib, so the source code and generated code can be provided under the same namespace without a merging step. - MyPy and PyLint are now invoked once for the entire codebase. - dprint will be used to format TypeScript/json files in the future instead of the slower prettier (currently turned off to avoid causing conflicts). It can automatically defer to prettier when formatting Svelte files. - svelte-check is now used for typechecking our Svelte code, which revealed a few typing issues that went undetected with the old system. - The Jest unit tests now work on Windows as well. If you're upgrading from Bazel, updated usage instructions are in docs/development.md and docs/build.md. A summary of the changes: - please remove node_modules and .bazel - install rustup (https://rustup.rs/) - install rsync if not already installed (on windows, use pacman - see docs/windows.md) - install Ninja (unzip from https://github.com/ninja-build/ninja/releases/tag/v1.11.1 and place on your path, or from your distro/homebrew if it's 1.10+) - update .vscode/settings.json from .vscode.dist
2022-11-27 06:24:20 +01:00
"pbkdf2",
2022-11-30 03:32:27 +01:00
"sha1",
"time",
2022-11-30 03:32:27 +01:00
"zstd 0.11.2+zstd.1.5.2",
]
Backups (#1685) * Add zstd dep * Implement backend backup with zstd * Implement backup thinning * Write backup meta * Use new file ending anki21b * Asynchronously backup on collection close in Rust * Revert "Add zstd dep" This reverts commit 3fcb2141d2be15f907269d13275c41971431385c. * Add zstd again * Take backup col path from col struct * Fix formatting * Implement backup restoring on backend * Normalize restored media file names * Refactor `extract_legacy_data()` A bit cumbersome due to borrowing rules. * Refactor * Make thinning calendar-based and gradual * Consider last kept backups of previous stages * Import full apkgs and colpkgs with backend * Expose new backup settings * Test `BackupThinner` and make it deterministic * Mark backup_path when closing optional * Delete leaky timer * Add progress updates for restoring media * Write restored collection to tempfile first * Do collection compression in the background thread This has us currently storing an uncompressed and compressed copy of the collection in memory (not ideal), but means the collection can be closed without waiting for compression to complete. On a large collection, this takes a close and reopen from about 0.55s to about 0.07s. The old backup code for comparison: about 0.35s for compression off, about 8.5s for zip compression. * Use multithreading in zstd compression On my system, this reduces the compression time of a large collection from about 0.55s to 0.08s. * Stream compressed collection data into zip file * Tweak backup explanation + Fix incorrect tab order for ignore accents option * Decouple restoring backup and full import In the first case, no profile is opened, unless the new collection succeeds to load. In the second case, either the old collection is reloaded or the new one is loaded. * Fix number gap in Progress message * Don't revert backup when media fails but report it * Tweak error flow * Remove native BackupLimits enum * Fix type annotation * Add thinning test for whole year * Satisfy linter * Await async backup to finish * Move restart disclaimer out of backup tab Should be visible regardless of the current tab. * Write restored collection in chunks * Refactor * Write media in chunks and refactor * Log error if removing file fails * join_backup_task -> await_backup_completion * Refactor backup.rs * Refactor backup meta and collection extraction * Fix wrong error being returned * Call sync_all() on new collection * Add ImportError * Store logger in Backend, instead of creating one on demand init_backend() accepts a Logger rather than a log file, to allow other callers to customize the logger if they wish. In the future we may want to explore using the tracing crate as an alternative; it's a bit more ergonomic, as a logger doesn't need to be passed around, and it plays more nicely with async code. * Sync file contents prior to rename; sync folder after rename. * Limit backup creation to once per 30 min * Use zstd::stream::copy_decode * Make importing abortable * Don't revert if backup media is aborted * Set throttle implicitly * Change force flag to minimum_backup_interval * Don't attempt to open folders on Windows * Join last backup thread before starting new one Also refactor. * Disable auto sync and backup when restoring again * Force backup on full download * Include the reason why a media file import failed, and the file path - Introduce a FileIoError that contains a string representation of the underlying I/O error, and an associated path. There are a few places in the code where we're currently manually including the filename in a custom error message, and this is a step towards a more consistent approach (but we may be better served with a more general approach in the future similar to Anyhow's .context()) - Move the error message into importing.ftl, as it's a bit neater when error messages live in the same file as the rest of the messages associated with some functionality. * Fix importing of media files * Minor wording tweaks * Save an allocation I18n strings with replacements are already strings, so we can skip the extra allocation. Not that it matters here at all. * Terminate import if file missing from archive If a third-party tool is creating invalid archives, the user should know about it. This should be rare, so I did not attempt to make it translatable. * Skip multithreaded compression on small collections Co-authored-by: Damien Elmes <gpg@ankiweb.net>
2022-03-07 06:11:31 +01:00
[[package]]
name = "zstd"
2022-09-24 04:39:21 +02:00
version = "0.11.2+zstd.1.5.2"
Backups (#1685) * Add zstd dep * Implement backend backup with zstd * Implement backup thinning * Write backup meta * Use new file ending anki21b * Asynchronously backup on collection close in Rust * Revert "Add zstd dep" This reverts commit 3fcb2141d2be15f907269d13275c41971431385c. * Add zstd again * Take backup col path from col struct * Fix formatting * Implement backup restoring on backend * Normalize restored media file names * Refactor `extract_legacy_data()` A bit cumbersome due to borrowing rules. * Refactor * Make thinning calendar-based and gradual * Consider last kept backups of previous stages * Import full apkgs and colpkgs with backend * Expose new backup settings * Test `BackupThinner` and make it deterministic * Mark backup_path when closing optional * Delete leaky timer * Add progress updates for restoring media * Write restored collection to tempfile first * Do collection compression in the background thread This has us currently storing an uncompressed and compressed copy of the collection in memory (not ideal), but means the collection can be closed without waiting for compression to complete. On a large collection, this takes a close and reopen from about 0.55s to about 0.07s. The old backup code for comparison: about 0.35s for compression off, about 8.5s for zip compression. * Use multithreading in zstd compression On my system, this reduces the compression time of a large collection from about 0.55s to 0.08s. * Stream compressed collection data into zip file * Tweak backup explanation + Fix incorrect tab order for ignore accents option * Decouple restoring backup and full import In the first case, no profile is opened, unless the new collection succeeds to load. In the second case, either the old collection is reloaded or the new one is loaded. * Fix number gap in Progress message * Don't revert backup when media fails but report it * Tweak error flow * Remove native BackupLimits enum * Fix type annotation * Add thinning test for whole year * Satisfy linter * Await async backup to finish * Move restart disclaimer out of backup tab Should be visible regardless of the current tab. * Write restored collection in chunks * Refactor * Write media in chunks and refactor * Log error if removing file fails * join_backup_task -> await_backup_completion * Refactor backup.rs * Refactor backup meta and collection extraction * Fix wrong error being returned * Call sync_all() on new collection * Add ImportError * Store logger in Backend, instead of creating one on demand init_backend() accepts a Logger rather than a log file, to allow other callers to customize the logger if they wish. In the future we may want to explore using the tracing crate as an alternative; it's a bit more ergonomic, as a logger doesn't need to be passed around, and it plays more nicely with async code. * Sync file contents prior to rename; sync folder after rename. * Limit backup creation to once per 30 min * Use zstd::stream::copy_decode * Make importing abortable * Don't revert if backup media is aborted * Set throttle implicitly * Change force flag to minimum_backup_interval * Don't attempt to open folders on Windows * Join last backup thread before starting new one Also refactor. * Disable auto sync and backup when restoring again * Force backup on full download * Include the reason why a media file import failed, and the file path - Introduce a FileIoError that contains a string representation of the underlying I/O error, and an associated path. There are a few places in the code where we're currently manually including the filename in a custom error message, and this is a step towards a more consistent approach (but we may be better served with a more general approach in the future similar to Anyhow's .context()) - Move the error message into importing.ftl, as it's a bit neater when error messages live in the same file as the rest of the messages associated with some functionality. * Fix importing of media files * Minor wording tweaks * Save an allocation I18n strings with replacements are already strings, so we can skip the extra allocation. Not that it matters here at all. * Terminate import if file missing from archive If a third-party tool is creating invalid archives, the user should know about it. This should be rare, so I did not attempt to make it translatable. * Skip multithreaded compression on small collections Co-authored-by: Damien Elmes <gpg@ankiweb.net>
2022-03-07 06:11:31 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
2022-09-24 04:39:21 +02:00
checksum = "20cc960326ece64f010d2d2107537f26dc589a6573a316bd5b1dba685fa5fde4"
Backups (#1685) * Add zstd dep * Implement backend backup with zstd * Implement backup thinning * Write backup meta * Use new file ending anki21b * Asynchronously backup on collection close in Rust * Revert "Add zstd dep" This reverts commit 3fcb2141d2be15f907269d13275c41971431385c. * Add zstd again * Take backup col path from col struct * Fix formatting * Implement backup restoring on backend * Normalize restored media file names * Refactor `extract_legacy_data()` A bit cumbersome due to borrowing rules. * Refactor * Make thinning calendar-based and gradual * Consider last kept backups of previous stages * Import full apkgs and colpkgs with backend * Expose new backup settings * Test `BackupThinner` and make it deterministic * Mark backup_path when closing optional * Delete leaky timer * Add progress updates for restoring media * Write restored collection to tempfile first * Do collection compression in the background thread This has us currently storing an uncompressed and compressed copy of the collection in memory (not ideal), but means the collection can be closed without waiting for compression to complete. On a large collection, this takes a close and reopen from about 0.55s to about 0.07s. The old backup code for comparison: about 0.35s for compression off, about 8.5s for zip compression. * Use multithreading in zstd compression On my system, this reduces the compression time of a large collection from about 0.55s to 0.08s. * Stream compressed collection data into zip file * Tweak backup explanation + Fix incorrect tab order for ignore accents option * Decouple restoring backup and full import In the first case, no profile is opened, unless the new collection succeeds to load. In the second case, either the old collection is reloaded or the new one is loaded. * Fix number gap in Progress message * Don't revert backup when media fails but report it * Tweak error flow * Remove native BackupLimits enum * Fix type annotation * Add thinning test for whole year * Satisfy linter * Await async backup to finish * Move restart disclaimer out of backup tab Should be visible regardless of the current tab. * Write restored collection in chunks * Refactor * Write media in chunks and refactor * Log error if removing file fails * join_backup_task -> await_backup_completion * Refactor backup.rs * Refactor backup meta and collection extraction * Fix wrong error being returned * Call sync_all() on new collection * Add ImportError * Store logger in Backend, instead of creating one on demand init_backend() accepts a Logger rather than a log file, to allow other callers to customize the logger if they wish. In the future we may want to explore using the tracing crate as an alternative; it's a bit more ergonomic, as a logger doesn't need to be passed around, and it plays more nicely with async code. * Sync file contents prior to rename; sync folder after rename. * Limit backup creation to once per 30 min * Use zstd::stream::copy_decode * Make importing abortable * Don't revert if backup media is aborted * Set throttle implicitly * Change force flag to minimum_backup_interval * Don't attempt to open folders on Windows * Join last backup thread before starting new one Also refactor. * Disable auto sync and backup when restoring again * Force backup on full download * Include the reason why a media file import failed, and the file path - Introduce a FileIoError that contains a string representation of the underlying I/O error, and an associated path. There are a few places in the code where we're currently manually including the filename in a custom error message, and this is a step towards a more consistent approach (but we may be better served with a more general approach in the future similar to Anyhow's .context()) - Move the error message into importing.ftl, as it's a bit neater when error messages live in the same file as the rest of the messages associated with some functionality. * Fix importing of media files * Minor wording tweaks * Save an allocation I18n strings with replacements are already strings, so we can skip the extra allocation. Not that it matters here at all. * Terminate import if file missing from archive If a third-party tool is creating invalid archives, the user should know about it. This should be rare, so I did not attempt to make it translatable. * Skip multithreaded compression on small collections Co-authored-by: Damien Elmes <gpg@ankiweb.net>
2022-03-07 06:11:31 +01:00
dependencies = [
2022-11-30 03:32:27 +01:00
"zstd-safe 5.0.2+zstd.1.5.2",
]
[[package]]
name = "zstd"
version = "0.12.3+zstd.1.5.2"
2022-11-30 03:32:27 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "76eea132fb024e0e13fd9c2f5d5d595d8a967aa72382ac2f9d39fcc95afd0806"
2022-11-30 03:32:27 +01:00
dependencies = [
"zstd-safe 6.0.4+zstd.1.5.4",
Backups (#1685) * Add zstd dep * Implement backend backup with zstd * Implement backup thinning * Write backup meta * Use new file ending anki21b * Asynchronously backup on collection close in Rust * Revert "Add zstd dep" This reverts commit 3fcb2141d2be15f907269d13275c41971431385c. * Add zstd again * Take backup col path from col struct * Fix formatting * Implement backup restoring on backend * Normalize restored media file names * Refactor `extract_legacy_data()` A bit cumbersome due to borrowing rules. * Refactor * Make thinning calendar-based and gradual * Consider last kept backups of previous stages * Import full apkgs and colpkgs with backend * Expose new backup settings * Test `BackupThinner` and make it deterministic * Mark backup_path when closing optional * Delete leaky timer * Add progress updates for restoring media * Write restored collection to tempfile first * Do collection compression in the background thread This has us currently storing an uncompressed and compressed copy of the collection in memory (not ideal), but means the collection can be closed without waiting for compression to complete. On a large collection, this takes a close and reopen from about 0.55s to about 0.07s. The old backup code for comparison: about 0.35s for compression off, about 8.5s for zip compression. * Use multithreading in zstd compression On my system, this reduces the compression time of a large collection from about 0.55s to 0.08s. * Stream compressed collection data into zip file * Tweak backup explanation + Fix incorrect tab order for ignore accents option * Decouple restoring backup and full import In the first case, no profile is opened, unless the new collection succeeds to load. In the second case, either the old collection is reloaded or the new one is loaded. * Fix number gap in Progress message * Don't revert backup when media fails but report it * Tweak error flow * Remove native BackupLimits enum * Fix type annotation * Add thinning test for whole year * Satisfy linter * Await async backup to finish * Move restart disclaimer out of backup tab Should be visible regardless of the current tab. * Write restored collection in chunks * Refactor * Write media in chunks and refactor * Log error if removing file fails * join_backup_task -> await_backup_completion * Refactor backup.rs * Refactor backup meta and collection extraction * Fix wrong error being returned * Call sync_all() on new collection * Add ImportError * Store logger in Backend, instead of creating one on demand init_backend() accepts a Logger rather than a log file, to allow other callers to customize the logger if they wish. In the future we may want to explore using the tracing crate as an alternative; it's a bit more ergonomic, as a logger doesn't need to be passed around, and it plays more nicely with async code. * Sync file contents prior to rename; sync folder after rename. * Limit backup creation to once per 30 min * Use zstd::stream::copy_decode * Make importing abortable * Don't revert if backup media is aborted * Set throttle implicitly * Change force flag to minimum_backup_interval * Don't attempt to open folders on Windows * Join last backup thread before starting new one Also refactor. * Disable auto sync and backup when restoring again * Force backup on full download * Include the reason why a media file import failed, and the file path - Introduce a FileIoError that contains a string representation of the underlying I/O error, and an associated path. There are a few places in the code where we're currently manually including the filename in a custom error message, and this is a step towards a more consistent approach (but we may be better served with a more general approach in the future similar to Anyhow's .context()) - Move the error message into importing.ftl, as it's a bit neater when error messages live in the same file as the rest of the messages associated with some functionality. * Fix importing of media files * Minor wording tweaks * Save an allocation I18n strings with replacements are already strings, so we can skip the extra allocation. Not that it matters here at all. * Terminate import if file missing from archive If a third-party tool is creating invalid archives, the user should know about it. This should be rare, so I did not attempt to make it translatable. * Skip multithreaded compression on small collections Co-authored-by: Damien Elmes <gpg@ankiweb.net>
2022-03-07 06:11:31 +01:00
]
[[package]]
name = "zstd-safe"
2022-09-24 04:39:21 +02:00
version = "5.0.2+zstd.1.5.2"
Backups (#1685) * Add zstd dep * Implement backend backup with zstd * Implement backup thinning * Write backup meta * Use new file ending anki21b * Asynchronously backup on collection close in Rust * Revert "Add zstd dep" This reverts commit 3fcb2141d2be15f907269d13275c41971431385c. * Add zstd again * Take backup col path from col struct * Fix formatting * Implement backup restoring on backend * Normalize restored media file names * Refactor `extract_legacy_data()` A bit cumbersome due to borrowing rules. * Refactor * Make thinning calendar-based and gradual * Consider last kept backups of previous stages * Import full apkgs and colpkgs with backend * Expose new backup settings * Test `BackupThinner` and make it deterministic * Mark backup_path when closing optional * Delete leaky timer * Add progress updates for restoring media * Write restored collection to tempfile first * Do collection compression in the background thread This has us currently storing an uncompressed and compressed copy of the collection in memory (not ideal), but means the collection can be closed without waiting for compression to complete. On a large collection, this takes a close and reopen from about 0.55s to about 0.07s. The old backup code for comparison: about 0.35s for compression off, about 8.5s for zip compression. * Use multithreading in zstd compression On my system, this reduces the compression time of a large collection from about 0.55s to 0.08s. * Stream compressed collection data into zip file * Tweak backup explanation + Fix incorrect tab order for ignore accents option * Decouple restoring backup and full import In the first case, no profile is opened, unless the new collection succeeds to load. In the second case, either the old collection is reloaded or the new one is loaded. * Fix number gap in Progress message * Don't revert backup when media fails but report it * Tweak error flow * Remove native BackupLimits enum * Fix type annotation * Add thinning test for whole year * Satisfy linter * Await async backup to finish * Move restart disclaimer out of backup tab Should be visible regardless of the current tab. * Write restored collection in chunks * Refactor * Write media in chunks and refactor * Log error if removing file fails * join_backup_task -> await_backup_completion * Refactor backup.rs * Refactor backup meta and collection extraction * Fix wrong error being returned * Call sync_all() on new collection * Add ImportError * Store logger in Backend, instead of creating one on demand init_backend() accepts a Logger rather than a log file, to allow other callers to customize the logger if they wish. In the future we may want to explore using the tracing crate as an alternative; it's a bit more ergonomic, as a logger doesn't need to be passed around, and it plays more nicely with async code. * Sync file contents prior to rename; sync folder after rename. * Limit backup creation to once per 30 min * Use zstd::stream::copy_decode * Make importing abortable * Don't revert if backup media is aborted * Set throttle implicitly * Change force flag to minimum_backup_interval * Don't attempt to open folders on Windows * Join last backup thread before starting new one Also refactor. * Disable auto sync and backup when restoring again * Force backup on full download * Include the reason why a media file import failed, and the file path - Introduce a FileIoError that contains a string representation of the underlying I/O error, and an associated path. There are a few places in the code where we're currently manually including the filename in a custom error message, and this is a step towards a more consistent approach (but we may be better served with a more general approach in the future similar to Anyhow's .context()) - Move the error message into importing.ftl, as it's a bit neater when error messages live in the same file as the rest of the messages associated with some functionality. * Fix importing of media files * Minor wording tweaks * Save an allocation I18n strings with replacements are already strings, so we can skip the extra allocation. Not that it matters here at all. * Terminate import if file missing from archive If a third-party tool is creating invalid archives, the user should know about it. This should be rare, so I did not attempt to make it translatable. * Skip multithreaded compression on small collections Co-authored-by: Damien Elmes <gpg@ankiweb.net>
2022-03-07 06:11:31 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
2022-09-24 04:39:21 +02:00
checksum = "1d2a5585e04f9eea4b2a3d1eca508c4dee9592a89ef6f450c11719da0726f4db"
Backups (#1685) * Add zstd dep * Implement backend backup with zstd * Implement backup thinning * Write backup meta * Use new file ending anki21b * Asynchronously backup on collection close in Rust * Revert "Add zstd dep" This reverts commit 3fcb2141d2be15f907269d13275c41971431385c. * Add zstd again * Take backup col path from col struct * Fix formatting * Implement backup restoring on backend * Normalize restored media file names * Refactor `extract_legacy_data()` A bit cumbersome due to borrowing rules. * Refactor * Make thinning calendar-based and gradual * Consider last kept backups of previous stages * Import full apkgs and colpkgs with backend * Expose new backup settings * Test `BackupThinner` and make it deterministic * Mark backup_path when closing optional * Delete leaky timer * Add progress updates for restoring media * Write restored collection to tempfile first * Do collection compression in the background thread This has us currently storing an uncompressed and compressed copy of the collection in memory (not ideal), but means the collection can be closed without waiting for compression to complete. On a large collection, this takes a close and reopen from about 0.55s to about 0.07s. The old backup code for comparison: about 0.35s for compression off, about 8.5s for zip compression. * Use multithreading in zstd compression On my system, this reduces the compression time of a large collection from about 0.55s to 0.08s. * Stream compressed collection data into zip file * Tweak backup explanation + Fix incorrect tab order for ignore accents option * Decouple restoring backup and full import In the first case, no profile is opened, unless the new collection succeeds to load. In the second case, either the old collection is reloaded or the new one is loaded. * Fix number gap in Progress message * Don't revert backup when media fails but report it * Tweak error flow * Remove native BackupLimits enum * Fix type annotation * Add thinning test for whole year * Satisfy linter * Await async backup to finish * Move restart disclaimer out of backup tab Should be visible regardless of the current tab. * Write restored collection in chunks * Refactor * Write media in chunks and refactor * Log error if removing file fails * join_backup_task -> await_backup_completion * Refactor backup.rs * Refactor backup meta and collection extraction * Fix wrong error being returned * Call sync_all() on new collection * Add ImportError * Store logger in Backend, instead of creating one on demand init_backend() accepts a Logger rather than a log file, to allow other callers to customize the logger if they wish. In the future we may want to explore using the tracing crate as an alternative; it's a bit more ergonomic, as a logger doesn't need to be passed around, and it plays more nicely with async code. * Sync file contents prior to rename; sync folder after rename. * Limit backup creation to once per 30 min * Use zstd::stream::copy_decode * Make importing abortable * Don't revert if backup media is aborted * Set throttle implicitly * Change force flag to minimum_backup_interval * Don't attempt to open folders on Windows * Join last backup thread before starting new one Also refactor. * Disable auto sync and backup when restoring again * Force backup on full download * Include the reason why a media file import failed, and the file path - Introduce a FileIoError that contains a string representation of the underlying I/O error, and an associated path. There are a few places in the code where we're currently manually including the filename in a custom error message, and this is a step towards a more consistent approach (but we may be better served with a more general approach in the future similar to Anyhow's .context()) - Move the error message into importing.ftl, as it's a bit neater when error messages live in the same file as the rest of the messages associated with some functionality. * Fix importing of media files * Minor wording tweaks * Save an allocation I18n strings with replacements are already strings, so we can skip the extra allocation. Not that it matters here at all. * Terminate import if file missing from archive If a third-party tool is creating invalid archives, the user should know about it. This should be rare, so I did not attempt to make it translatable. * Skip multithreaded compression on small collections Co-authored-by: Damien Elmes <gpg@ankiweb.net>
2022-03-07 06:11:31 +01:00
dependencies = [
"libc",
"zstd-sys",
]
2022-11-30 03:32:27 +01:00
[[package]]
name = "zstd-safe"
version = "6.0.4+zstd.1.5.4"
2022-11-30 03:32:27 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "7afb4b54b8910cf5447638cb54bf4e8a65cbedd783af98b98c62ffe91f185543"
2022-11-30 03:32:27 +01:00
dependencies = [
"libc",
"zstd-sys",
]
Backups (#1685) * Add zstd dep * Implement backend backup with zstd * Implement backup thinning * Write backup meta * Use new file ending anki21b * Asynchronously backup on collection close in Rust * Revert "Add zstd dep" This reverts commit 3fcb2141d2be15f907269d13275c41971431385c. * Add zstd again * Take backup col path from col struct * Fix formatting * Implement backup restoring on backend * Normalize restored media file names * Refactor `extract_legacy_data()` A bit cumbersome due to borrowing rules. * Refactor * Make thinning calendar-based and gradual * Consider last kept backups of previous stages * Import full apkgs and colpkgs with backend * Expose new backup settings * Test `BackupThinner` and make it deterministic * Mark backup_path when closing optional * Delete leaky timer * Add progress updates for restoring media * Write restored collection to tempfile first * Do collection compression in the background thread This has us currently storing an uncompressed and compressed copy of the collection in memory (not ideal), but means the collection can be closed without waiting for compression to complete. On a large collection, this takes a close and reopen from about 0.55s to about 0.07s. The old backup code for comparison: about 0.35s for compression off, about 8.5s for zip compression. * Use multithreading in zstd compression On my system, this reduces the compression time of a large collection from about 0.55s to 0.08s. * Stream compressed collection data into zip file * Tweak backup explanation + Fix incorrect tab order for ignore accents option * Decouple restoring backup and full import In the first case, no profile is opened, unless the new collection succeeds to load. In the second case, either the old collection is reloaded or the new one is loaded. * Fix number gap in Progress message * Don't revert backup when media fails but report it * Tweak error flow * Remove native BackupLimits enum * Fix type annotation * Add thinning test for whole year * Satisfy linter * Await async backup to finish * Move restart disclaimer out of backup tab Should be visible regardless of the current tab. * Write restored collection in chunks * Refactor * Write media in chunks and refactor * Log error if removing file fails * join_backup_task -> await_backup_completion * Refactor backup.rs * Refactor backup meta and collection extraction * Fix wrong error being returned * Call sync_all() on new collection * Add ImportError * Store logger in Backend, instead of creating one on demand init_backend() accepts a Logger rather than a log file, to allow other callers to customize the logger if they wish. In the future we may want to explore using the tracing crate as an alternative; it's a bit more ergonomic, as a logger doesn't need to be passed around, and it plays more nicely with async code. * Sync file contents prior to rename; sync folder after rename. * Limit backup creation to once per 30 min * Use zstd::stream::copy_decode * Make importing abortable * Don't revert if backup media is aborted * Set throttle implicitly * Change force flag to minimum_backup_interval * Don't attempt to open folders on Windows * Join last backup thread before starting new one Also refactor. * Disable auto sync and backup when restoring again * Force backup on full download * Include the reason why a media file import failed, and the file path - Introduce a FileIoError that contains a string representation of the underlying I/O error, and an associated path. There are a few places in the code where we're currently manually including the filename in a custom error message, and this is a step towards a more consistent approach (but we may be better served with a more general approach in the future similar to Anyhow's .context()) - Move the error message into importing.ftl, as it's a bit neater when error messages live in the same file as the rest of the messages associated with some functionality. * Fix importing of media files * Minor wording tweaks * Save an allocation I18n strings with replacements are already strings, so we can skip the extra allocation. Not that it matters here at all. * Terminate import if file missing from archive If a third-party tool is creating invalid archives, the user should know about it. This should be rare, so I did not attempt to make it translatable. * Skip multithreaded compression on small collections Co-authored-by: Damien Elmes <gpg@ankiweb.net>
2022-03-07 06:11:31 +01:00
[[package]]
name = "zstd-sys"
version = "2.0.7+zstd.1.5.4"
Backups (#1685) * Add zstd dep * Implement backend backup with zstd * Implement backup thinning * Write backup meta * Use new file ending anki21b * Asynchronously backup on collection close in Rust * Revert "Add zstd dep" This reverts commit 3fcb2141d2be15f907269d13275c41971431385c. * Add zstd again * Take backup col path from col struct * Fix formatting * Implement backup restoring on backend * Normalize restored media file names * Refactor `extract_legacy_data()` A bit cumbersome due to borrowing rules. * Refactor * Make thinning calendar-based and gradual * Consider last kept backups of previous stages * Import full apkgs and colpkgs with backend * Expose new backup settings * Test `BackupThinner` and make it deterministic * Mark backup_path when closing optional * Delete leaky timer * Add progress updates for restoring media * Write restored collection to tempfile first * Do collection compression in the background thread This has us currently storing an uncompressed and compressed copy of the collection in memory (not ideal), but means the collection can be closed without waiting for compression to complete. On a large collection, this takes a close and reopen from about 0.55s to about 0.07s. The old backup code for comparison: about 0.35s for compression off, about 8.5s for zip compression. * Use multithreading in zstd compression On my system, this reduces the compression time of a large collection from about 0.55s to 0.08s. * Stream compressed collection data into zip file * Tweak backup explanation + Fix incorrect tab order for ignore accents option * Decouple restoring backup and full import In the first case, no profile is opened, unless the new collection succeeds to load. In the second case, either the old collection is reloaded or the new one is loaded. * Fix number gap in Progress message * Don't revert backup when media fails but report it * Tweak error flow * Remove native BackupLimits enum * Fix type annotation * Add thinning test for whole year * Satisfy linter * Await async backup to finish * Move restart disclaimer out of backup tab Should be visible regardless of the current tab. * Write restored collection in chunks * Refactor * Write media in chunks and refactor * Log error if removing file fails * join_backup_task -> await_backup_completion * Refactor backup.rs * Refactor backup meta and collection extraction * Fix wrong error being returned * Call sync_all() on new collection * Add ImportError * Store logger in Backend, instead of creating one on demand init_backend() accepts a Logger rather than a log file, to allow other callers to customize the logger if they wish. In the future we may want to explore using the tracing crate as an alternative; it's a bit more ergonomic, as a logger doesn't need to be passed around, and it plays more nicely with async code. * Sync file contents prior to rename; sync folder after rename. * Limit backup creation to once per 30 min * Use zstd::stream::copy_decode * Make importing abortable * Don't revert if backup media is aborted * Set throttle implicitly * Change force flag to minimum_backup_interval * Don't attempt to open folders on Windows * Join last backup thread before starting new one Also refactor. * Disable auto sync and backup when restoring again * Force backup on full download * Include the reason why a media file import failed, and the file path - Introduce a FileIoError that contains a string representation of the underlying I/O error, and an associated path. There are a few places in the code where we're currently manually including the filename in a custom error message, and this is a step towards a more consistent approach (but we may be better served with a more general approach in the future similar to Anyhow's .context()) - Move the error message into importing.ftl, as it's a bit neater when error messages live in the same file as the rest of the messages associated with some functionality. * Fix importing of media files * Minor wording tweaks * Save an allocation I18n strings with replacements are already strings, so we can skip the extra allocation. Not that it matters here at all. * Terminate import if file missing from archive If a third-party tool is creating invalid archives, the user should know about it. This should be rare, so I did not attempt to make it translatable. * Skip multithreaded compression on small collections Co-authored-by: Damien Elmes <gpg@ankiweb.net>
2022-03-07 06:11:31 +01:00
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "94509c3ba2fe55294d752b79842c530ccfab760192521df74a081a78d2b3c7f5"
Backups (#1685) * Add zstd dep * Implement backend backup with zstd * Implement backup thinning * Write backup meta * Use new file ending anki21b * Asynchronously backup on collection close in Rust * Revert "Add zstd dep" This reverts commit 3fcb2141d2be15f907269d13275c41971431385c. * Add zstd again * Take backup col path from col struct * Fix formatting * Implement backup restoring on backend * Normalize restored media file names * Refactor `extract_legacy_data()` A bit cumbersome due to borrowing rules. * Refactor * Make thinning calendar-based and gradual * Consider last kept backups of previous stages * Import full apkgs and colpkgs with backend * Expose new backup settings * Test `BackupThinner` and make it deterministic * Mark backup_path when closing optional * Delete leaky timer * Add progress updates for restoring media * Write restored collection to tempfile first * Do collection compression in the background thread This has us currently storing an uncompressed and compressed copy of the collection in memory (not ideal), but means the collection can be closed without waiting for compression to complete. On a large collection, this takes a close and reopen from about 0.55s to about 0.07s. The old backup code for comparison: about 0.35s for compression off, about 8.5s for zip compression. * Use multithreading in zstd compression On my system, this reduces the compression time of a large collection from about 0.55s to 0.08s. * Stream compressed collection data into zip file * Tweak backup explanation + Fix incorrect tab order for ignore accents option * Decouple restoring backup and full import In the first case, no profile is opened, unless the new collection succeeds to load. In the second case, either the old collection is reloaded or the new one is loaded. * Fix number gap in Progress message * Don't revert backup when media fails but report it * Tweak error flow * Remove native BackupLimits enum * Fix type annotation * Add thinning test for whole year * Satisfy linter * Await async backup to finish * Move restart disclaimer out of backup tab Should be visible regardless of the current tab. * Write restored collection in chunks * Refactor * Write media in chunks and refactor * Log error if removing file fails * join_backup_task -> await_backup_completion * Refactor backup.rs * Refactor backup meta and collection extraction * Fix wrong error being returned * Call sync_all() on new collection * Add ImportError * Store logger in Backend, instead of creating one on demand init_backend() accepts a Logger rather than a log file, to allow other callers to customize the logger if they wish. In the future we may want to explore using the tracing crate as an alternative; it's a bit more ergonomic, as a logger doesn't need to be passed around, and it plays more nicely with async code. * Sync file contents prior to rename; sync folder after rename. * Limit backup creation to once per 30 min * Use zstd::stream::copy_decode * Make importing abortable * Don't revert if backup media is aborted * Set throttle implicitly * Change force flag to minimum_backup_interval * Don't attempt to open folders on Windows * Join last backup thread before starting new one Also refactor. * Disable auto sync and backup when restoring again * Force backup on full download * Include the reason why a media file import failed, and the file path - Introduce a FileIoError that contains a string representation of the underlying I/O error, and an associated path. There are a few places in the code where we're currently manually including the filename in a custom error message, and this is a step towards a more consistent approach (but we may be better served with a more general approach in the future similar to Anyhow's .context()) - Move the error message into importing.ftl, as it's a bit neater when error messages live in the same file as the rest of the messages associated with some functionality. * Fix importing of media files * Minor wording tweaks * Save an allocation I18n strings with replacements are already strings, so we can skip the extra allocation. Not that it matters here at all. * Terminate import if file missing from archive If a third-party tool is creating invalid archives, the user should know about it. This should be rare, so I did not attempt to make it translatable. * Skip multithreaded compression on small collections Co-authored-by: Damien Elmes <gpg@ankiweb.net>
2022-03-07 06:11:31 +01:00
dependencies = [
"cc",
"libc",
"pkg-config",
Backups (#1685) * Add zstd dep * Implement backend backup with zstd * Implement backup thinning * Write backup meta * Use new file ending anki21b * Asynchronously backup on collection close in Rust * Revert "Add zstd dep" This reverts commit 3fcb2141d2be15f907269d13275c41971431385c. * Add zstd again * Take backup col path from col struct * Fix formatting * Implement backup restoring on backend * Normalize restored media file names * Refactor `extract_legacy_data()` A bit cumbersome due to borrowing rules. * Refactor * Make thinning calendar-based and gradual * Consider last kept backups of previous stages * Import full apkgs and colpkgs with backend * Expose new backup settings * Test `BackupThinner` and make it deterministic * Mark backup_path when closing optional * Delete leaky timer * Add progress updates for restoring media * Write restored collection to tempfile first * Do collection compression in the background thread This has us currently storing an uncompressed and compressed copy of the collection in memory (not ideal), but means the collection can be closed without waiting for compression to complete. On a large collection, this takes a close and reopen from about 0.55s to about 0.07s. The old backup code for comparison: about 0.35s for compression off, about 8.5s for zip compression. * Use multithreading in zstd compression On my system, this reduces the compression time of a large collection from about 0.55s to 0.08s. * Stream compressed collection data into zip file * Tweak backup explanation + Fix incorrect tab order for ignore accents option * Decouple restoring backup and full import In the first case, no profile is opened, unless the new collection succeeds to load. In the second case, either the old collection is reloaded or the new one is loaded. * Fix number gap in Progress message * Don't revert backup when media fails but report it * Tweak error flow * Remove native BackupLimits enum * Fix type annotation * Add thinning test for whole year * Satisfy linter * Await async backup to finish * Move restart disclaimer out of backup tab Should be visible regardless of the current tab. * Write restored collection in chunks * Refactor * Write media in chunks and refactor * Log error if removing file fails * join_backup_task -> await_backup_completion * Refactor backup.rs * Refactor backup meta and collection extraction * Fix wrong error being returned * Call sync_all() on new collection * Add ImportError * Store logger in Backend, instead of creating one on demand init_backend() accepts a Logger rather than a log file, to allow other callers to customize the logger if they wish. In the future we may want to explore using the tracing crate as an alternative; it's a bit more ergonomic, as a logger doesn't need to be passed around, and it plays more nicely with async code. * Sync file contents prior to rename; sync folder after rename. * Limit backup creation to once per 30 min * Use zstd::stream::copy_decode * Make importing abortable * Don't revert if backup media is aborted * Set throttle implicitly * Change force flag to minimum_backup_interval * Don't attempt to open folders on Windows * Join last backup thread before starting new one Also refactor. * Disable auto sync and backup when restoring again * Force backup on full download * Include the reason why a media file import failed, and the file path - Introduce a FileIoError that contains a string representation of the underlying I/O error, and an associated path. There are a few places in the code where we're currently manually including the filename in a custom error message, and this is a step towards a more consistent approach (but we may be better served with a more general approach in the future similar to Anyhow's .context()) - Move the error message into importing.ftl, as it's a bit neater when error messages live in the same file as the rest of the messages associated with some functionality. * Fix importing of media files * Minor wording tweaks * Save an allocation I18n strings with replacements are already strings, so we can skip the extra allocation. Not that it matters here at all. * Terminate import if file missing from archive If a third-party tool is creating invalid archives, the user should know about it. This should be rare, so I did not attempt to make it translatable. * Skip multithreaded compression on small collections Co-authored-by: Damien Elmes <gpg@ankiweb.net>
2022-03-07 06:11:31 +01:00
]