🤖 Improve integration test diagnostics for flaky tool policy tests #46
Conversation
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
Instead of timing out silently, the tests now: - Wait for either stream-end OR stream-error - Log all collected events for debugging - Log the actual error message and error type if stream fails - Fail with a clear message pointing to the logs This will help us understand what's actually failing when the tests are flaky in CI (e.g., is the AI trying to use disabled tools? Is there an API error? etc.) _Generated with `cmux`_
Enhanced and to provide better error messages when tests fail: **waitForEvent improvements:** - When timing out, now automatically logs stream-error details if present - Shows error message and error type for debugging **assertStreamSuccess improvements:** - Replaced generic expect() calls with descriptive Error messages - Shows all collected events when any assertion fails - Distinguishes between different failure modes: - Stream didn't complete (no stream-end) - Stream errored (has stream-error) - Stream completed but missing final message - Includes the actual error message in the failure output This makes test failures in CI much easier to diagnose without needing to add test-specific logging everywhere. _Generated with `cmux`_
Now that waitForEvent and assertStreamSuccess provide detailed diagnostics, we can remove the test-specific logging code. The tests now: - Wait for either stream-end or stream-error (prevents timeout) - Rely on helper functions for error diagnostics - Are much more readable and maintainable All the diagnostic logging happens automatically in the helpers, making it a general fix for all integration tests, not just these two. _Generated with `cmux`_
ammario
force-pushed
the
fix-tool-policy-test-flakiness
branch
from
d2e6dcb to
d343a40
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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.
Problem
The tool policy integration tests are flaky in CI, timing out waiting for
stream-endevents with no useful diagnostic information. When they fail, we only see:This tells us nothing about what actually went wrong.
Solution
This PR makes general improvements to test helpers that benefit all integration tests:
1. Enhanced
waitForEventhelper2. Enhanced
assertStreamSuccesshelper3. Fixed tool policy tests to wait for completion
stream-endORstream-error(prevents timeout)Impact
These changes benefit all integration tests, not just tool policy tests. Any test using
waitForEventorassertStreamSuccesswill now provide much better error messages when failing.Next Steps
Once this PR is merged and we see the tests fail in CI, we'll get detailed error logs that will help us:
Generated with
cmux