Repository navigation
test(crew_log): run lease and real-crash suites on macOS again - #13898
Conversation
GPT 5.6 Review — ✅ no blocking findingsGPT 5.6 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: |
Design Review (Fable 5) — ✅ PASSDesign-level review of Design-Verdict: PASS Confirm this PR's own macOS backend shards ran the 14 un-skipped tests green before merge. [DESIGN-REVIEWED] ecd9502 |
Opus 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: |
|
macOS evidence for #10704 (head
The shards run |
Problem / Motivation
Two crew-log test suites never run on macOS.
test/test_crew_log_lease.pyandtest/test_crew_log_real_crash.pyskip ondarwin"tracked in #10704".Why it matters
These are the only tests that kill a real writer or hold a lease from a second process. On macOS a regression in crash repair or lease ownership would ship unseen.
What changed (motivation → approach → change)
The darwin skip was a bisect step. It isolated these suites from a macOS shard failing with
ENOENT: .kirocrew-deps.lockin the apps deps-provisioning tests.That race is fixed on main by #10464 (9173117), which creates the dependency lock with
O_CREAT|O_EXCL. The skip has nothing left to isolate.Both skip conditions drop
darwinand keepwin32, whose lock and SIGKILL semantics really differ. The bisect comments go with the skip.Backwards compatibility
Compatible: test-only change; no production code touched.
Tests
No new tests. The 14 tests in the two files now run on the macOS backend shards. Linux run passes locally (14 passed).
Manual verification
N/A — the CI macOS backend shard is the verification; its job is linked in a comment.
Related Issues
Closes #10704