(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
* PEP8 dbproxy.py
* PEP8 errors.py
* PEP8 httpclient.py
* PEP8 lang.py
* PEP8 latex.py
* Add decorator to deprectate key words
* Make replacement for deprecated attribute optional
* Use new helper `_print_replacement_warning()`
* PEP8 media.py
* PEP8 rsbackend.py
* PEP8 sound.py
* PEP8 stdmodels.py
* PEP8 storage.py
* PEP8 sync.py
* PEP8 tags.py
* PEP8 template.py
* PEP8 types.py
* Fix DeprecatedNamesMixinForModule
The class methods need to be overridden with instance methods, so every
module has its own dicts.
* Use `# pylint: disable=invalid-name` instead of id
* PEP8 utils.py
* Only decorate `__getattr__` with `@no_type_check`
* Fix mypy issue with snakecase
Importing it from `anki._vendor` raises attribute errors.
* Format
* Remove inheritance of DeprecatedNamesMixin
There's almost no shared code now and overriding classmethods with
instance methods raises mypy issues.
* Fix traceback frames of deprecation warnings
* remove fn/TimedLog (dae)
Neither Anki nor add-ons appear to have been using it
* fix some issues with stringcase use (dae)
- the wheel was depending on the PyPI version instead of our vendored
version
- _vendor:stringcase should not have been listed in the anki py_library.
We already include the sources in py_srcs, and need to refer to them
directly. By listing _vendor:stringcase as well, we were making a
top-level stringcase library available, which would have only worked for
distributing because the wheel definition was also incorrect.
- mypy errors are what caused me to mistakenly add the above - they
were because the type: ignore at the top of stringcase.py was causing
mypy to completely ignore the file, so it was not aware of any attributes
it contained.
Will allow importing the Protobuf without pulling in the rest of
the library. This is not a full PEP420 namespace, and the wheel still
bundles everything - it just makes things easier in a Bazel workspace.
I originally tried with PEP420, but it required more invasive changes,
and I ran into issues with mypy.
The existing code was really difficult to reason about:
- The default notetype depended on the selected deck, and vice versa,
and this logic was buried in the deck and notetype choosing screens,
and models.py.
- Changes to the notetype were not passed back directly, but were fired
via a hook, which changed any screen in the app that had a notetype
selector.
It also wasn't great for performance, as the most recent deck and tags
were embedded in the notetype, which can be expensive to save and sync
for large notetypes.
To address these points:
- The current deck for a notetype, and notetype for a deck, are now
stored in separate config variables, instead of directly in the deck
or notetype. These are cheap to read and write, and we'll be able to
sync them individually in the future once config syncing is updated in
the future. I seem to recall some users not wanting the tag saving
behaviour, so I've dropped that for now, but if people end up missing
it, it would be simple to add as an extra auxiliary config variable.
- The logic for getting the starting deck and notetype has been moved
into the backend. It should be the same as the older Python code, with
one exception: when "change deck depending on notetype" is enabled in
the preferences, it will start with the current notetype ("curModel"),
instead of first trying to get a deck-specific notetype.
- ModelChooser has been duplicated into notetypechooser.py, and it
has been updated to solely be concerned with keeping track of a selected
notetype - it no longer alters global state.
- anki._backend stores the protobuf files and rsbackend.py code
- pylib modules import protobuf messages directly from the
_pb2 files, and explicitly export any will be returned or consumed
by public pylib functions, so that calling code can import from pylib
- the "rsbackend" no longer imports and re-exports protobuf messages
- pylib can just consume them directly.
- move errors to errors.py
Still todo:
- rsbridge
- finishing the work on rsbackend, and check what we need to add
back to the original file location to avoid breaking add-ons
- notetypes are fetched from the DB as needed, and cached in Python
- handle note type changes in the backend. Multiple operations can now
be performed in one go, but this is not currently exposed in the GUI.
- extra methods to grab sorted note type names quickly, and fetch by
name
- col.models.save() without a provided notetype is now a no-op
- note loading/saving handled in the backend
- notes with no valid cards can now be added
- templates can now be deleted even if they would previously
orphan notes
a number of fixmes have been left in notes.py and models.py
- all .ftl files for a language are concatenated into a single file
at build time
- all languages are included in the binary
- external ftl files placed in the ftl folder can override the
built-in definitions
- constants are automatically generated for each string key
- dropped the separate StringsGroup enum
The parsing step is considerably slower in Python, but if parsing
is moved out of the test function, Python wins at 45ms to Rust's 67ms
on 10,000 rounds, presumably due to the overhead of serializing to
Protobuf. Not enough of a difference to justify the inclusion of extra
dependencies and duplicating the lookup code in any case.
We can now show replay buttons for the audio contained in {{FrontSide}}
without having to play it again when the answer is shown.
The template code now always defers FrontSide rendering, as it wasn't
a big saving, and meant the logic had to be implemented twice.
