This allows us to use `uv` for:
1. Installing a pinned Python version
2. Installing the dependency packages using `uv`'s pip compatible interface.
4. Bootstrapping `mach` without a Python installion on the host, using `uv
run`
This change also introduces a new 'composite' GitHub action to setup
python in the different CI workflows. There is no support for externally
managed python installations and virtual environments. These could be
added in the future.
Fixes#34095, #34547
Signed-off-by: Mukilan Thiyagarajan <mukilan@igalia.com>
* Bump bindgen to 0.71.1
Signed-off-by: Jonathan Schwender <schwenderjonathan@gmail.com>
* Set SCCACHE_GHA_VERSION to allow purging cache
changing the version allows us to purge the cache.
Signed-off-by: Jonathan Schwender <schwenderjonathan@gmail.com>
* android CI: Enable sccache again
Signed-off-by: Jonathan Schwender <schwenderjonathan@gmail.com>
---------
Signed-off-by: Jonathan Schwender <schwenderjonathan@gmail.com>
This patch switches servo to use `uv` for both installing a pinned
Python version as well as installing the dependency packages using
`uv`'s pip compatible interface. It also introduces a new 'composite'
GitHub action to setup python in the different CI workflows.
There is no support for externally managed python installations and
virtual environments. These could be added in the future.
Fixes#34095
Signed-off-by: Mukilan Thiyagarajan <mukilan@igalia.com>
wptrunner breaks macos builds because it doesn't work well with 3.13 due
to two issues:
1. The current version (10.3.0) of the 'pillow' dependency of wptrunner
breaks with >=3.13 and needs to be upgraded.
2. Python 3.13 has removed the 'cgi' module which was deprecated in 3.11
and has no direct replacement. There are two files in wptrunner that
use the cgi module and one of them is the vendored 'html5lib'.
As a quick fix, pin the Python version on MacOS runner until
web-platform-tests/wpt#48585 is addressed.
Signed-off-by: Mukilan Thiyagarajan <mukilan@igalia.com>
This [issue][1] was reported by GitHub user @RedYetiDev via the Security
Advisory reporting mechanism on GitHub. The fix is also based on their
proposed solution.
The issue is that `refs/pull/{pr_number}/head` points to the latest
commit of a PR and so it could be different than the commit that was
reviewed when the try label was applied. The fix is to use the exact commit
sha at the point when the try job is triggered, which is available in
the `github` context as `github.event.pull_request.head.sha`.
[1]: https://github.com/servo/servo/security/advisories/GHSA-fxqr-xgh8-3577
Signed-off-by: Mukilan Thiyagarajan <mukilan@igalia.com>
Co-authored-by: Aviv Keller <38299977+RedYetiDev@users.noreply.github.com>
* Make `crown` optional
Add the optional `--use-crown` flag to mach
* --use-crown for all platforms in CI
Signed-off-by: Jonathan Schwender <jonathan.schwender@huawei.com>
* Add documentation for `--use-crown`
Signed-off-by: Jonathan Schwender <jonathan.schwender@huawei.com>
* Update python/servo/command_base.py
Co-authored-by: Mukilan Thiyagarajan <mukilanthiagarajan@gmail.com>
Signed-off-by: Jonathan Schwender <schwenderjonathan@gmail.com>
* Raise Error if CARGO_BUILD_RUSTC conflicts with --use-crown
Signed-off-by: Jonathan Schwender <schwenderjonathan@gmail.com>
* add dummy RUSTFLAG to trigger re-checking
Signed-off-by: Jonathan Schwender <schwenderjonathan@gmail.com>
---------
Signed-off-by: Jonathan Schwender <jonathan.schwender@huawei.com>
Signed-off-by: Jonathan Schwender <schwenderjonathan@gmail.com>
Co-authored-by: Mukilan Thiyagarajan <mukilanthiagarajan@gmail.com>
Use the `cargo-install` action to cache the cargo-deny output.
`cargo-deny` is currently unconditionally installed during bootstrap,
and takes around 2 minutes to install, so caching should give a
significant speedup
Signed-off-by: Jonathan Schwender <jonathan.schwender@huawei.com>
This was previously disabled in #30508 due to sccache not
working well with crown. The sccache issue (mozilla/sccache#861)
linked in that PR is now closed and [testing][1] on my fork also
seems to indicated we should be able to turn on sccache again.
[1]: 9154196647
Signed-off-by: Mukilan Thiyagarajan <mukilan@igalia.com>
Thus the build will immediately fail if a PR modifies Cargo.toml but
forgets to include the changes in Cargo.lock
This was previously checked by lockfile_changed.sh after building
normally, wasting resources.
* ci: Switch to version 4 of GitHub artifact actions
This switches to version 4 of the GitHub artifact actions, which
requires producing a single artifact per job and adding merge steps. In
addition, the names of artifacts are standardized:
- Build: <profile>-binary-<platform>
- Full WPT results (only on failure): wpt-full-logs-<platform>-<layout>
- Filtered WPT results (only on failure): wpt-filtered-logs-<platform>-<layout>
* Delete merged build timings and combine with Result job
* Always archives logs even after test failures
* Correct the name of the log files for WPT import
It seems that timing issues (related to MacOS or the GitHub MacOS)
runners can sometimes cause `hdiutil detach` to fail. Instead of having
this cause the entire build to fail, fail gracefully. This is
essentially a non-issue as the CI environment is always cleaned up when
using GitHub Actions.
Fixes#30757.
Make it so that all try builds go through try.yml and pass
workflow_call arguments as expected to subsequent workflows.
Co-authored-by: Samson <16504129+sagudev@users.noreply.github.com>
* Replace virtualenv with Python's built-in venv.
* Apply Delan's suggestions and make a couple small fixes
- Fix a tidy warning about directories that don't exist
- Use shutil instead of the redundant get_exec_path
- Miscellaneous cleanups
* Fix typo in environment variable
* fix bug where pip still tries to the wrong site-packages
---------
Co-authored-by: Martin Robinson <mrobinson@igalia.com>
Co-authored-by: Delan Azabani <dazabani@igalia.com>
* Remove script_plugins
* Use crown instead of script_plugins
* crown_is_not_used
* Use crown in command base
* bootstrap crown
* tidy happy
* disable sccache
* Bring crown in tree
* Install crown from tree
* fix windows ci
* fix warning
* fix mac
libscript_plugins.dylib is not available anymore
* Update components/script/lib.rs
Co-authored-by: Martin Robinson <mrobinson@igalia.com>
* Update for nightly-2023-03-18
Mostly just based off https://github.com/servo/servo/pull/30630
* Always install crown
it's slow only when there is new version
* Run crown test with `mach test-unit`
* Small fixups; better trace_in_no_trace tests
* Better doc
* crown in config.toml
* Fix tidy for real
* no sccache on rustc_wrapper
* document rustc overrides
* fixup of compiletest
* Make a few minor comment adjustments
* Fix a typo in python/servo/platform/base.py
Co-authored-by: Samson <16504129+sagudev@users.noreply.github.com>
* Proper test types
* Ignore tidy on crown/tests
---------
Co-authored-by: Martin Robinson <mrobinson@igalia.com>
This is an attempt to fix errors on the Mac CI when running `hdiutil`
that look like this:
```
Run python3 ./mach package --release
hdiutil: create failed - Resource busy
Creating Servo.app
Copying files
Swapping prefs
Finding dylibs and relinking
Adding version to Credits.rtf
Creating dmg
Packaging MacOS dmg exited with return value 1
Error: Process completed with exit code 1.
```
This approach was taken from
https://github.com/actions/runner-images/issues/7522.
Triggering from labels means that we have less actions running and less
false job failures spamming project members. Plus, we have more
flexibility with labels rather than the backward compatibility we have
set up for bors comments.
* Run main and try jobs with debug assertions
* use single quotes in workflow expressions
* set force-debug-assertions in main.yml
* set force-debug-assertions as part of decision job
* fix typo in MachCommands.build
* fix more hardcoded profile names
* fix tidy
* split cargo_profile_option on windows
* Fix running servoshell and unit tests through a symlink
* rename steps to make them less confusing
* fix more hardcoded cargo profile options
* fix missing inputs in linux-wpt and mac-wpt
* make filename an inherent method of Resource
* rework release-with-debug-assertions profile to production profile
* rework resource logic to eliminate std_test_override
* set production flag in nightly release builds
* clean up servobuild.example and windows.yml
* oops forgot to check in embedder_traits/build.rs
* fix mach test-unit behaviour through symlink
* unit tests only need current_dir and ancestors
* fix macOS package smoketest breakage
* expect css/css-color/currentcolor-003 to crash under layout 2013
* fix more references to {force,release-with}-debug-assertions
* fix local build failures under --profile production
This change adds and alternate method for triggering try changes.
Instead of comments, changes are triggered via applying labels to pull
requests. The action will remove the label from the request and start
the requested jobs.
This will require creating at least a few labels:
- T-full
- T-linux-wpt-2013
- T-linux-wpt-2020
- T-macos
- T-windows
More labels can be added as we support more configurations.
The good thing about this change is that try jobs against the actual
branch in the pull request instead of the master branch. This means
that changes to CI can be tested (unlike for comment processing).
One bit caveat with this change is that when adding multiple labels, a
CI job is triggered for each. Only one real build will run for each
label, but whether or more try jobs is triggered is a race condition.
The first CI job to successfully remove the label will actually trigger
the job. If the same job removes two compatible labels, then they can
share a build (for instance two types of WPT linux jobs). If not there
will be two. Note that this is at least as efficient as the current
behavior.
There are currently two ways to run try. One is to push to the `try` or
`try-*` branches and the other is to trigger a workflow via GitHub
comment. This change combines these methods into one workflow. In
addition, WPT results are reported together rather than separately and
filtered results for all WPT tests are bundled together in the same
artifact.
There were some issues with the way that the `--release` and `--dev`
arguments were handled in mach commands.
- Not all commands accepted them in the same way. For instance `./mach
test-wpt` didn't really accept them at all.
- If you did not pass either of them, mach would try to guess which
build you meant. This guess was often quite surprising as it wasn't
printed and it depended on the state of the your target directory,
which is difficult to remember.
- The `dev` profile is colloquially called a "debug" profile and some
commands accepted `-d` or `--debug...` like arguments, but `--debug`
with `./mach run` meant run in a debugger. It was easy to mix this
up.
This change:
- Centralizes where build type argument processing happens. Now it the
same shared decorator in CommandBase.
- Uses a `BuildType` enum instead of passing around two different
booleans. This reduces the error checking for situations where both
are true.
- Be much less clever about guessing what build to use. Now if you
don't specify a build type, `--dev` is chosen. I think this behavior
matches cargo.
- Makes it so that `./mach test-wpt` accepts the exact same arguments
and has the same behavior as other commands. In addition, the suite
correct for `test-wpt` is removed. There are only two suites now and
it's quite unlikely that people will confuse WPT tests for rust unit
tests.
Now that the new version of GStreamer fixes this issue, we can remove
the workarounds for this problem as well as all of the homebrew
bootstrapping logic.
There is currently no way to trigger unit tests using the bors-servo try
infrastructure. This change makes it so that using try-<platform>
triggers unit tests for that platform. This will be especially useful
when testing changes for Windows before trying to land them.
