scheduler: use per-group average duration for unknown tests#114
scheduler: use per-group average duration for unknown tests#114DanverImbue merged 4 commits intomainfrom
Conversation
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Instead of a flat 1-second default, compute the average duration of tests in each group from historical data and use that as the fallback for tests without individual timing history. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
jsk11235
force-pushed
the
worktree-scheduler-defaults
branch
from
f5dad40 to
faf7ba6
Compare
![github-actions[bot]](https://avatars.githubusercontent.com/in/15368?s=60&v=4){kind=small}
| @@ -211,7 +218,7 @@ mod tests { | |||
| durations.insert("medium_test".to_string(), Duration::from_secs(5)); | |||
| durations.insert("fast_test".to_string(), Duration::from_secs(1)); | |||
|
|
|||
There was a problem hiding this comment.
[test_coverage] (severity 3/5) (confidence 0.92)
The diff adds per-group average duration logic but no test verifies this behavior. All existing scheduler tests pass empty HashMap::new() for group_defaults, so the per-group fallback path (where a group has a computed average that differs from the 1s default) is never tested. A test should be added where some tests have historical durations and others in the same group don't, verifying the group average is used instead of the 1s fallback.
There was a problem hiding this comment.
Vet found 2 issues.
[commit_message_mismatch] (severity 2/5) (confidence 0.85)
The diff includes changes to skills/offload-onboard/SKILL.md (Step 9 rewrite with baseline comparison instructions) that are unrelated to the user request about replacing flat default_duration with per-group average in the scheduler. This is an unauthorized change not mentioned in the request.
| @@ -211,7 +218,7 @@ mod tests { | |||
| durations.insert("medium_test".to_string(), Duration::from_secs(5)); | |||
| durations.insert("fast_test".to_string(), Duration::from_secs(1)); | |||
|
|
|||
There was a problem hiding this comment.
[test_coverage] (severity 3/5) (confidence 0.92)
No test was added to verify the per-group average duration behavior. There should be a test where some tests in a group have historical durations and others don't, verifying that the unknown tests use the group average rather than the 1s fallback. The existing test test_schedule_uses_default_for_unknown still passes an empty group_to_default_duration map, so it only tests the 1s fallback path, not the new per-group average feature.
3 participants
Summary
group()accessor toTestInstancedurationsmap and passes them toschedule_lptTest plan
cargo fmt --checkpassescargo clippy --all-targets --all-featurespassescargo nextest runpasses (128/128 tests)🤖 Generated with Claude Code