Repository navigation
ci(v1-gate): flag holds that need a maintainer - #1132
Conversation
From 2026-11-09 the gate held v1 for about 36 hours because no VM could boot on ubuntu-26.04 (#1123). The hold issue looked like a test failure ("failed with no failing scenario recorded"), was offered to contributors as a good first issue, and never reached a maintainer. Two holds now say so and are parked for a person: - v1_gate_compare.py reports infrastructure=true when no failed suite produced test results on either side (the machines never booted); - the hold issue has been open for more than 24 hours. In both cases the comment starts with "Needs a maintainer", and the issue gets needs-human and priority/critical-urgent and loses good first issue / help wanted. Ordinary holds are unchanged. Refs #1123 Assisted-by: Claude Opus 5.5 via Claude Code Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Signed-off-by: Danathar <Danathar@users.noreply.github.com>
|
Codecov Report✅ All modified and coverable lines are covered by tests. 📢 Thoughts on this report? Let us know! |
1c6e9bd
ℹ️ Sentinel notice — informational; author is trusted (owner), merge is not blockedHive flagged this PR (author @Danathar, head
Hive did not add the |
There was a problem hiding this comment.
HUMAN DECISION NEEDED — @Danathar: confirm workflow correctness and trust-boundary safety before relying on this incomplete review.
Summary: This change directs stalled release checks to maintainers instead of contributors. I found no verified defect in the changes I could read, but could not complete checks of the surrounding automation after GitHub access reached its hourly limit.
What I need from you: Have a maintainer complete the integration and security review.
Reviewed revision 1c6e9bde4e33923c22b2c15d3ce2f13cd4734b2b: PR body, complete diff, and existing comments/reviews. Correctness remains unverified around the new classification (scripts/v1_gate_compare.py:125-140) and its workflow consumer (.github/workflows/update-v1-tag.yml:241-245). Security remains unverified around the new issue-editing operations (.github/workflows/update-v1-tag.yml:311-318): surrounding triggers, permissions, and input sources were unavailable. No local checkout was located; tests were not independently run. These are review limits, not confirmed defects.
No findings from: intent-alignment, style, docs-currency.
Confidence: 3/5 (needs attention) — 2 perspectives requires a human decision
— hive: agent=reviewer backend=copilot model=gpt-6.1-sol copilot=1.0.88
Summary: from 2026-11-09 the
v1gate heldv1for about 36 hours because no test VM could boot onubuntu-26.04(#1123, fixed by #1131). The gate did open its hold issue, but it read like a test failure ("failed with no failing scenario recorded"). The hive offered it to contributors as a "good first issue", and it never reached a maintainer. This PR makes two kinds of hold say plainly that they need a maintainer and park them for a person. Ordinary holds don't change.When a hold is flagged for a maintainer
results.json) on either side, the merged tests or thev1baseline.scripts/v1_gate_compare.pynow printsinfrastructure=truefor this case. If even one suite produced results, it's an ordinary hold.What happens then
needs-humanandpriority/critical-urgent, andgood first issue/help wantedare removed. In the hive,needs-humanon an issue parks it for a person, so it shows in the maintainers' queue instead of being offered to contributors.Files
scripts/v1_gate_compare.py:nothing_ran(), theinfrastructureoutput, and a summary line.tests/unit/test_v1_gate_compare.py: 5 new tests (both sides empty, one side has results, failed scenarios, one of two suites has results, no artifacts at all). The CLI test now expects both output lines..github/workflows/update-v1-tag.yml: the escalation in "Move or hold v1".docs/skills/ci-ops/ops/references/candidate-vm-validation.md: describes it.Checks run locally
The 14 tests in
test_v1_gate_compare.pypass (run directly; pytest runs in CI).ruff check tests/ scripts/v1_gate_compare.py --select E,F,W --ignore E501: clean.actionlint: no findings, same as
main.The "Move or hold v1" script, run with a stubbed
ghin three cases:All three exit 1 as before.
Refs #1123
🤖 Generated with Claude Code