ci(v1.x): sync the #91 cancellation migration - #220
Conversation
Content matches fleet-cancellation-migration (#219) — byte-identical to main once that merges. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01S94XethbGXpAb4DRKMD4kt
|
Note Gemini is unable to generate a review for this pull request due to the file types involved not being currently supported. |
|
2 blockers found. Prior 2 finding(s) remain. 1. Missing events in opt-in gate (Claude)File: .github/workflows/claude-review.yml:36 if: ${{ (github.event.action == 'labeled' && github.event.label.name == 'claude-review') || (github.event.action != 'labeled' && (vars.CLAUDE_ALWAYS_ON == 'true' || contains(github.event.pull_request.labels.*.name, 'claude-review'))) }}2. Missing events in opt-in gate (Gemini)File: .github/workflows/gemini-review.yml:44 if: ${{ (github.event.action == 'labeled' && github.event.label.name == 'gemini-review') || (github.event.action != 'labeled' && (vars.GEMINI_ALWAYS_ON == 'true' || contains(github.event.pull_request.labels.*.name, 'gemini-review'))) }} |
Review feedback on oauth#219 — consistency with the PR's permissions- ceiling goal, applied fleet-wide. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01S94XethbGXpAb4DRKMD4kt
The retained note claiming the label is matched only in the reusable contradicted the caller gate two lines below (it now names the label); reworded to state the deliberate duplication. Narrating comments that restated the adjacent declaration are dropped. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01S94XethbGXpAb4DRKMD4kt
|
Codex review: no blockers; byte-identical port of #219. One finding: a review-lane concurrency race (an older push's stale review can cancel a newer one), tracked upstream as ai-review-prompts#92 and likely behind the flaky duplicate review runs. Not a release issue; fine to land with #92 as follow-up. |
Kris's precondition for this migration (harper#2385 thread, ai-review-prompts#92/#94): stale runs are refused admission to the review job's cancelling group and self-fail if superseded after acquiring the slot, so removing the caller-side cancellation cannot let an older run cancel a newer admitted review. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01S94XethbGXpAb4DRKMD4kt
![github-actions[bot]](https://avatars.githubusercontent.com/in/15368?s=60&v=4){kind=small}
heskew
left a comment
heskew
left a comment
There was a problem hiding this comment.
Double-check at c646b7e: no blockers.
All five changed caller workflows are byte-for-byte identical to current main after #219. The ai-review-prompts #94 pin is exact and consistent across uses/inputs; permission ceilings, label/ALWAYS_ON policy, pre-queue freshness gating, reusable-owned cancellation, and the post-acquire stale-head check all line up. I honored the two resolved label-gate threads as the accepted cost policy.
The red Claude review job is historical: it ran before #219 merged and failed because this caller did not yet match the default branch. That prerequisite is now satisfied, so a rerun should clear it. Node 22/24, Bun, validation, and security checks are green.
1 participant
Brings the reusable-owned-cancellation migration (#219) to
v1.x— byte-identical to main once #219 merges, so merge #219 first, then this (caller validation compares against the default branch).🤖 Generated with Claude Code
https://claude.ai/code/session_01S94XethbGXpAb4DRKMD4kt