Repository navigation
🐛 tracker: merge outcomes always render in the Merge gate column; pending CI is not 'red' - #15408
Conversation
…ding CI is not 'red'
A merge outcome (merging / merge-failed) on a row the lane set back to
060 was drawn in CI / Review because stageTrackerAnchor mapped stage 60
to column 50 regardless of merge_gate; the chip carried GitHub's raw
error text (a PUT …/merge URL) and the caption read 'CI is red
(check-pending)' and 'fell back to stage/060-reviewed'.
- stageTrackerAnchor: a merge_gate state of merging / merge-failed
anchors the row in the Merge gate column whatever its numeric stage.
- Go: stageTrackerMergeRefusal translates GitHub's refusal into a kind,
a plain sentence and (for 405 'required status checks are expected')
the check count; the SPA renders a fixed short chip ('Waiting for CI ·
4 checks', 'Merge failed · retrying', 'Merge failed · branch
protection') with the raw text only after 'Details:' in the tooltip;
.ham-gate-detail is clamped with an ellipsis.
- stageTrackerCIStateWords: check-/status- tokens read 'CI running',
'CI green', 'CI failed: <check>' — pending is never red (Go caption
and SPA gate words).
- stageTrackerStageName: cues and tooltips name a stage in words
('back to Reviewed'), not by its slug — a minimal local fix; the copy
sweep retires stage slugs globally.
- automerge: the sweep remembers GitHub's 'required status checks are
expected' 405 per head and skips the merge call as check-pending while
the PR is still blocked, instead of re-PUTting /merge every pass.
Fixes #15404
Signed-off-by: Andrew Anderson <andy@clubanderson.com>
|
[APPROVALNOTIFIER] This PR is NOT APPROVED This pull-request has been approved by: The full list of commands accepted by this bot can be found here. DetailsNeeds approval from an approver in each of these files:Approvers can indicate their approval by writing |
Open PR overlap checkBase branch: Conflicts are checked with every PR merged onto the current tip of
|
|
@clubanderson is auto-promoted for contributor-lane merging on this PR: the last 5 merged PRs by this author in this hive's repositories have no revert or sentinel-alert evidence. CI, reviewer verdict, diff fences, soak, holds and branch protection still apply. |
|
@clubanderson this PR is held at Failing checks:
Next: the hive does not push to your branch. Push a new head (or reply here); CI and the review swarm re-run automatically and the PR returns to the normal path. |
|
stage 080-awaiting-merge-admission → 071-needs-author: required-check-failing:build-and-test |
…d-column node suite Signed-off-by: Andrew Anderson <andy@clubanderson.com>
|
stage 080-awaiting-merge-admission → 050-review-pending: review-gate:no-review-bundle |
There was a problem hiding this comment.
Style / maintainability review (head e4b72c8)
1. src/docs/dashboard-sections.md:1161-1162 — the new paragraph splits the legend table. The blank line + prose inserted before the | 110 | **Merged** … row ends the Markdown table, so that row renders as a stray literal | 110 | ... line. Move the paragraph below the table. While there, the 060 row (line 1153) still quotes the old tooltip wording CI still running (build-and-test), which this PR changes to CI running (build-and-test) (stage_tracker_merge_gate_test.go:213).
| any stage after the merge call drew *N of M required status checks are expected* (`merge_failed_kind: checks-pending`, `merge_failed_checks`) | **Waiting for CI** | *4 checks* | ⟳ while the sweep is on it, else 🕓 — *… It merges once they pass.* |
-
-A merge outcome (**Merging**, **Merge failed**, **Waiting for CI**) always lights the **Merge gate** column, whatever numeric stage the lane fell back to (hivecommons/docs#512 drew "Merge failed" in CI / Review because the PR sat at 060).
| 110 | **Merged** (the ✓ cap) | — | — |
+
+A merge outcome (**Merging**, **Merge failed**, **Waiting for CI**) always lights the **Merge gate** column, whatever numeric stage the lane fell back to (hivecommons/docs#512 drew "Merge failed" in CI / Review because the PR sat at 060).2. static-src/js/tracker-words.js:652-659 — a second, already-divergent refusal classifier. stageTrackerMergeRefusalFromRaw re-implements Go's stageTrackerMergeRefusal (stage_tracker_merge_refusal.go:48-85) "abridged": it has no review / method / merge-queue kinds, so 405 At least 1 approving review is required is review server-side but protection client-side. The only writer of merge_failed sets merge_failed_kind in the same statement (stage_tracker_merge_gate.go:184-185), and the SPA ships inside the same binary, so the "older payload" branch is unreachable in production — it only exists so fixtures without merge_failed_kind (stage_tracker_merging_hold_test.go:295, stage_tracker_merge_failed_column_test.go cases 3-7) keep passing. Suggest deleting the JS fallback and giving those fixtures merge_failed_kind/merge_failed_text, so there is one source of truth for the mapping.
3. static-src/js/tracker-words.js:690-692 — stageTrackerMergeFailedShort is now dead. Its last caller was the removed merge-failed branch; nothing in static-src/js references it or STAGE_TRACKER_MERGE_FAILED_SHORT_MAX any more (only stage_tracker_merge_gate_test.go:314 and stage_tracker_merging_hold_test.go:473-475). The rewritten comment "(tooltips and toasts; …)" describes callers that do not exist. Remove the helper, the const and their asserts, or at least drop the comment claim.
— hive: agent=ci-maintainer backend=copilot model=claude-fable-5.1 copilot=1.0.88
There was a problem hiding this comment.
Reviewed (intent-alignment) — the diff covers every bullet of #15404 (column anchor, pending≠red, retrying vs failed, raw error out of the chip, "Waiting for CI · N checks", stage-slug leak, sweep memory); all five stageTrackerAnchor callers pass item, and the 10 m MergeAttemptRetention bounds the forced Merge-gate anchor. No in-scope findings.
Out of scope (backlog, not blocking): the queued-merge path src/pkg/github/automerge/automerge_sweep.go:2681-2698 still has no requiredChecksExpected branch — a 405 "N of M required status checks are expected" there goes straight to refusedMerge → "merge-failed" and re-PUTs /merge each pass, the same loop this PR closes at the other three sites. The PR body says "three merge sites", so this is a known gap rather than a regression.
— hive: agent=adjudicator-3 backend=copilot model=claude-fable-5.1 copilot=1.0.88
|
Hive filed review findings that do not block this PR as backlog items:
Cap: 10 backlog item(s) per PR per day. |
There was a problem hiding this comment.
Reviewed (correctness) — no in-scope findings at e4b72c8. Column anchoring, refusal translation, humanizeStageReason ordering (the check-runs/status-check cases still precede stageTrackerCIStateWords, stage_tracker.go:1322-1328) and the refusal-memory locking (checksExpectedKnown runs under mergeRefusalsMu in knownMergeRefusal, automerge_merge_refusal.go:168-172) all check out; the node harness exercises the exact docs#512 payload.
Adjacent, not blocking (backlog): the queued lane's merge site (trySweepQueuedPR, automerge_sweep.go:2681-2700) is the one merge path not wrapped in requiredChecksExpected, so a 405 "required status checks are expected" there still falls through refusedMerge → "merge-failed" and is re-PUT every pass. Pre-existing; the issue's report was the self-authored lane.
— hive: agent=adjudicator backend=copilot model=claude-fable-5.1 copilot=1.0.88
|
@clubanderson this PR is held at Why:
Next: the hive does not push to your branch. Push a new head (or reply here); CI and the review swarm re-run automatically and the PR returns to the normal path. |
|
Thank you for your contribution! Your PR has been merged. We'd love to hear how your experience was: share feedback |
Fixes #15404
Operator report (hive's own spoke, row docs#496 → PR hivecommons/docs#512, 2026-10-29 11:04 ET): the amber "⟳ Merge failed · merge sweep running ⬡" chip sat in the CI / Review column, its label spilled GitHub's raw
PUT https://api.github.com/…error, and the caption read "fell back to stage/060-reviewed — … CI is red (check-pending)".Root causes
stageTrackerAnchormapped stage 60 → column 50 unconditionally and the gate chip is drawn on whichever segment is active, so a merge attempt on a row the lane set back to 060 landed in CI / Review./merge, and each refusal was recorded as a failed merge attempt that lit "Merge failed".Changes
static-src, regeneratedindex.html):stageTrackerAnchor(stage, stages, kind, item)— amerge_gate.stateofmerging/merge-failedanchors the row in the Merge gate column whatever its numeric stage (a stale gate on a PR back at ≤ 030 does not jump it forward). The merge-failed branch now precedessweep_running: "Merge failed · retrying" while the sweep is on it, else "Merge failed · "; a 405 "required checks expected" refusal is honestly "Waiting for CI · 4 checks". The chip is a fixed short phrase (≤ 5 words, no URL/verb) frommerge_failed_kind, with the plain sentence and GitHub's raw text (after "Details:") in the tooltip;.ham-gate-detailgetsmax-width+ ellipsis. Gate words: pending → "CI running", failing → "CI failed: lint".stageTrackerStageNamegives cues/tooltips plain words ("fell back to Reviewed") — a minimal local fix in the same helper; the copy agent is retiring stage slugs globally.stageTrackerMergeRefusal(newstage_tracker_merge_refusal.go) translates refusals → kind / sentence / check count (merge_failed_kind,merge_failed_text,merge_failed_checksonmerge_gate);stageTrackerCIStateWordsrenderscheck-*/status-*tokens as "CI running" / "CI green" / "CI failed: " inhumanizeStageReason.Engine.requiredChecksExpectedwrapsrequiredChecksExpectedMergeErrorat the three merge sites and remembers the head;knownMergeRefusalreturnscheck-pendingon later passes while the same head is stillblocked(re-armed by a new head, an unblocked state or the 30 m recheck).TestStageTrackerMergeFailedColumnNodewith the exact payload (stage 60,merge-failed,sweep_running, raw 405) — column 80, chip wording, nohttp/PUT, ≤ 5 words, Details line, CI words, cue text; GoTestStageTrackerMergeRefusal,TestHumanizeStageReasoncases,TestRequiredChecksExpectedRemembered.Notes