From 1d55d1599657742b1f8b76621cf756b99ddd8529 Mon Sep 17 00:00:00 2001 From: Mao Nakamoto <41178744+maonakamoto@users.noreply.github.com> Date: Sat, 29 Aug 2026 06:09:01 +0200 Subject: [PATCH] chore(ci): finish the catomean -> bitbaum rename in auto-merge-sweep PR #227 moved the deploy workflow to the bitbaum org but left two incident notes in auto-merge-sweep.sh pointing at catomean/ repo paths that no longer resolve. The comments are the only record of why the sweep skips red bases and merges oldest-first, so a dead path there costs a future reader the receipt. Found because local main had diverged: it carried this half of the rename in a commit whose deploy.yml half shipped as #227. Co-Authored-By: Claude Opus 5 (1M context) Claude-Session: https://claude.ai/code/session_01TDhA9vfTasuixhGQ2NVmcE --- scripts/ci/auto-merge-sweep.sh | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/scripts/ci/auto-merge-sweep.sh b/scripts/ci/auto-merge-sweep.sh index 174a2939..0e57a273 100644 --- a/scripts/ci/auto-merge-sweep.sh +++ b/scripts/ci/auto-merge-sweep.sh @@ -93,7 +93,7 @@ else # base quietly collecting more of them and getting harder to diagnose. But # when the PR *is* the repair, the same rule deadlocks the repo — the fix # cannot travel the path its own redness blocks, and only a human can move - # it. Seen in catomean/aoz-housing on 2026-08-07: E2E red on the base, the + # it. Seen in bitbaum/aoz-housing on 2026-08-07: E2E red on the base, the # fix sitting green in a PR, every sweep refusing politely. # # So identify WHICH jobs are red and let a PR through only if its own checks @@ -128,7 +128,7 @@ merged_any=0 # OLDEST FIRST. `gh pr list` returns newest-first, and this loop merges the # first eligible PR and stops — so the newest green PR wins every sweep and an -# older one can wait indefinitely. Observed in catomean/fleetcrown on +# older one can wait indefinitely. Observed in bitbaum/fleetcrown on # 2026-08-06: two consecutive sweeps merged the two newest PRs while three # older green ones were never even evaluated. With several agent sessions # opening PRs continuously, "newest wins" is starvation, and it starves the PR