Repository navigation
fix(evals): pass the eval self-tests on native Windows - #4553
Merged
ScriptedAlchemy merged 2 commits intoOct 26, 2026
Merged
Conversation
Report ambient-tokio findings with POSIX paths, write the adoption report as UTF-8, run shell fixtures under Git's MSYS bash instead of the System32 WSL launcher, and print build-eval-binary paths in a form native Windows callers can open. The real-model runner owns agents through POSIX process groups, so its test is incompatible on Windows like evals/hermetic:run_test. Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
Contributor
Author
|
I'll fix CI failures and address comments from users with write access. I'll skip comments containing "(aside)".
|
|
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Native Windows
bazel test --config=ci //...on393edb94failed four Python test targets. No open PR touches these files (#4543 changes only.bazelrc/MODULE.bazeltoolchain pins).//scripts:ambient_tokio_check_teststr(path.relative_to(root))reported findings ascrates\demo\....as_posix()//evals/agent_adoption:selftestprint/report.mdin cp1252 can't encodeΔ;["bash", ...]resolves toSystem32\bash.exe(the WSL launcher, since CreateProcess searches System32 before PATH);carry: copy is read-onlyexpected0o400, but on Windows the only mode is the read-only bit (0o444)bash()resolves Git Bash throughgit; read-only expectation per platform//scripts:build_eval_binary_testbash; Git'sbin/bash.exewrapper puts/usr/binahead of the fakeuname/bazel; fakes written with CRLF; the script printed MSYS paths (/tmp/...) that native callers can't openusr/bin/bash.exewith its dir behind the fakes, pinsunametoLinuxby default, and writes fakes with\n.build-eval-binary.shnow prints both branches throughabsolute()(pwd -Won MSYS, elsepwd), so re-entry withTRACEDECAY_BIN=<output>returns the same string//evals/memory:real_model_runner_testos.killpg/setsid(scripts/lib/portable_process.py). Python on Windows has neither, and the test timed out at 300starget_compatible_withmarks the test incompatible on Windows, the same way as//evals/hermetic:run_testOn Linux and macOS,
pwd -Wis an invalid option, soabsolute()falls back topwdand the output does not change.Verified on Windows Server (Bazel 9.3.1):
bazel test --config=cion the three compatible targets passes. The fourth is reported as incompatible. Each script also passes when run directly withpython.Link to Devin session: https://app.devin.ai/sessions/b7efb5226c9845d097483ac9b2e3d9c3
Open in Devin Desktop: https://app.devin.ai/desktop/session/b7efb5226c9845d097483ac9b2e3d9c3?variant=devin
Requested by: @ScriptedAlchemy