Ended · 15 subagents
A pull request whose branch was pushed on top of another's sometimes gets no CI and reports conflicts that git does not have
Seen on 2026-10-22 with #362 (branch 361-home-tab, branched from #360's branch, which then merged). The PR showed merge_base null. Pushes at 04:43 and 05:15 created no CI / pull_request run at all; the required CI sat at 'expected'. update_branch answered 'conflicting' on home.tsx, ground-truth.tsx and home.md, though main's only new commit was #360's merge commit, whose content the branch already had (git merge origin/main was clean). A local merge of main pushed at 05:26 got CI at once. #364, branched from main, got CI within seconds. Find why the pull_request trigger and the branch update misjudge the merge base for a branch stacked on a merged PR.
Work
Repositories and pull requests
On it
syntaqxClaude CodeReleased
since
Stacked-branch PR: merge base / merge ref / update_branch conflict and silent CI skip (services/work, services/repos, services/actions/trigger)
syntaqxClaude CodeReleased
since
Scope: services/work pull merge base / pull_request trigger / update_branch for branches stacked on a merged PR; Checks: cargo tests
Sessions
In Chat- s
Untitled sessionClaude CodeEnded
syntaqx· ended
- s
Untitled sessionClaude CodeEnded
syntaqx· ended
Why the work is the way it is, and how sure that is