Repository navigation
test(dashboard): wait on signals in the agent-switch publish test - #19506
Conversation
First Principles Review (Opus 5.5) — ⏭️ skippedRevision |
UX Review (Opus 5.5) — ⏭️ skippedRevision |
Design Review (Opus 5.5) — ✅ PASSDesign-level review of Design-Verdict: PASS Nothing to check. [DESIGN-REVIEWED] d2c0517 Description read[DESCRIPTION-READ] 176853066eb7b935067d8581330264558fd173327468472c02765f96be03294f That sha256 names everything this verdict read: the PR description, plus the 1 evidence file(s) the workflow collected from it and handed this review. It covers that evidence because the media strip replaces every attachment URL with the same placeholder, so an edit swapping one attachment for another would leave the description's captured text identical while what this review judged changed. Recompute it with |
Opus 5.5 Review — ✅ no blocking findingsReviewed Verdict parsed from the review's SHA-scoped output markers for commit False positive or not applicable? A repository writer can comment: |
GPT 6.1 Review — ✅ no blocking findingsGPT 6.1 completed its review of This comment is updated in place on each push. Review detailsNo findings. False positive or not applicable? A repository writer can comment: |
|
The red |
b71770b to
e00d7a3
Compare
…agent-switch publish test
e00d7a3 to
d2c0517
Compare
Problem / Motivation
Goal:
test_agent_and_kind_are_not_published_separatelypasses on a slow Windows runner instead of failing at random.On Windows CI the test fails with
CancelledError/TimeoutErroratasyncio.wait_for(entered_resolution.wait(), timeout=1). Linux on the same head is green.Why it matters
Every PR whose Windows backend shard picks up this test can go red for no reason. That costs reruns and hides real failures.
Not a goal
What changed (motivation approach change)
The test waited at most 1 s for the handler to reach the stubbed warm-up, then 1 s more for the response.
Before the warm-up the handler does real off-loop reads of the session record. On a busy Windows runner those can take longer than 1 s, so the wait ran out and the request was cancelled.
Both waits use
await_untilfromkiro_crew.testing.wait: it awaits the same event and task, bounded only by the lost-run limit (half the test timeout), and fails by name.test/wait_discipline_baseline.txtdrops this file from 3 to 1 short fixed waits.Backwards compatibility
Compatible: test-only change.
Removes nothing: only two waits in one test change shape; no product code is touched.
Tests
test_agent_and_kind_are_not_published_separatelykeeps its assertions: a sender crossing resolution sees the old agent and kind, then the new pair.Determinism proof (Windows, Python 3.12):
63c155a582TimeoutErrorManual verification
N/A - unit coverage sufficient; the change is inside one test.
Related Issues
Closes #18960
Pattern harvest
Not generalizable: the rule already exists (testing-conventions D1, enforced by
test/test_wait_discipline_ratchet.py); this lowers one more file's count.Checklist