Skip to content

test(ask_user): de-flake two CI-timing failures (unblocks #237)#238

Merged
AVADSA25 merged 1 commit into
mainfrom
claude/fix-askuser-flake
Jul 13, 2026
Merged

test(ask_user): de-flake two CI-timing failures (unblocks #237)#238
AVADSA25 merged 1 commit into
mainfrom
claude/fix-askuser-flake

Conversation

@AVADSA25

Copy link
Copy Markdown
Owner

Two pre-existing timing flakes (pass in isolation, unrelated to the ruff bump in #237) scoped to their own question-id / polled for the lagging write. Test-only. 10/10 stable, full suite 2497 passed.

…237)

PR #237 (a harmless ruff version bump) went red on CI/smoke with two failures
that have nothing to do with ruff — both pass in isolation and are pre-existing
timing flakes:

  * test_deadline_timeout: `assert len(timeouts) == 1` saw 2. ask() is
    synchronous and finalizes exactly one timeout per call, so the extra event
    came from a background-threaded ask() in a NEIGHBOURING test finalizing late
    into this test's audit log (_log_event reads the log path live at
    write-time). Now scoped: count only emit/timeout events whose
    extra.pending_question_id is THIS test's qid.
  * test_round_trip: "notification not written". ask() writes the pending
    record and the notification as two separate writes; the test polled for the
    record then asserted on the notification, which can lag under load. Now
    polls for the notification too.

Both are test-only changes; runtime code untouched. Verified: the two tests
pass 10/10 under repeated runs; full suite 2497 passed.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
@AVADSA25
AVADSA25 merged commit 7d09e8f into main Jul 13, 2026
1 check passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants