Repository navigation
🐛 Activity Monitor: person-owed PR steps light PR in progress, not Merge queue - #16593
Conversation
|
[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 |
…rge queue A PR parked at 085 because a person has to act on the PR itself (rerun CI, approve workflow runs, resolve a GitHub permission, answer review findings or red checks still open on its head) is not queued, merging or merge-ready. Its chevron now lights PR in progress — "Waiting for @who · rerun shared ci on GitHub", person icon kept — and the Merge queue column stays empty. Only rows actually queued / merging / merged, or a merge-ready PR (review approved, CI green) parked on a merge decision, sit in Merge queue. A bare 085 with no track evidence keeps the gate (#14821). - stageTrackerItemColumn (Go) / stageTrackerAnchor (JS): new stageTrackerPersonActsOnPR predicate, mirrored on both sides. - Unauthorized word/caption: "Waiting for @author · <step> on GitHub" (a hive / bot / clanker author yields "a maintainer"). - Nested PR rows carry unauthorized_step so the word matches. - Tests: Go table (#16524 shape and its twins) + rows fixture. Fixes #16592 Signed-off-by: Andrew Anderson <andy@clubanderson.com>
cbbaa63 to
7d4ee65
Compare
Open PR overlap checkBase branch: Conflicts are checked with every PR merged onto the current tip of
|
|
Hive status — updated in place; edits do not notify. Full history is in the Activity Monitor. @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. Review findings from Correctness review @ 7d4ee65 — the #16524 shape itself is handled correctly (unauthorized branch, 1. The "bare 085 keeps the gate (#14821)" carve-out is unreachable with served rows — JS: 2. (low) Go caption never names the author. PR body says the "word / caption reads 'Waiting for Advisory (no change requested): with the rule as written, a merge-ready 085 whose CI re-runs ( Hive filed review findings that do not block this PR as backlog items:
Cap: 10 backlog item(s) per PR per day. Review findings from Read the predicate against the tree at One small mismatch worth a look, not a blocker:
Open question, not a finding: |
Signed-off-by: Andrew Anderson <andy@clubanderson.com>
There was a problem hiding this comment.
Read stage_tracker_vocabulary.go, stage_tracker_build.go (where the tracks are populated), stage_tracker_activity.go, and the JS mirror at the PR head (72203c1). The #16524 case itself is handled correctly, but the "bare 085 keeps the gate" rule the predicate documents is not reachable in the served payload.
1. unknown is counted as not-merge-ready evidence, so every bare 085 PR leaves Merge queue (medium)
src/pkg/dashboard/stage_tracker_vocabulary.go:138-141 — stageTrackerNotMergeReadyEvidence returns true whenever a track on the current head has State != "approved" / State != "passing". That includes "unknown".
In production neither track is ever nil:
stage_tracker_build.go:463→stageTrackerCIActivity, which starts fromtrackerUnknownActivity(pr.HeadSHA, "CI unknown · …")(stage_tracker_activity.go:57) and returns it unless checks are observed;stage_tracker_build.go:475→stageTrackerReviewActivity, which starts fromtrackerUnknownActivity(pr.HeadSHA, "Review unknown · …")(stage_tracker_activity.go:436);stage_tracker_build.go:535/:550backfillunknownon any head mismatch.
Both carry Head = it.Head, so stageTrackerActivityOnHead is true. Result: a bare 085 PR with no unauthorized: reason — a hive/supersession-pending park ("Superseded by #N — decide", the #14821 case), a hold, any decision park whose head hive has not yet seen checks or a verdict for — renders in PR in progress and only moves back to Merge queue once review is approved and CI is passing. That is the opposite of the predicate's own comment (:103-104, "A bare 085 with no track evidence keeps the gate") and of the Rule in the PR body. The test that guards this (stage_tracker_item_column_test.go:170-172, "bare 085 decision keeps the gate (#14821)") sets both tracks to nil, a shape the builder never emits; with ci("unknown"), review("unknown") it would return trackerColumnPR.
The JS mirror has the same behaviour (tracker-row-anchor.js, stageTrackerPersonActsOnPR: ci.state !== 'passing').
Suggested fix, both sides: treat unknown (and empty) as absent evidence — e.g. known := a != nil && a.State != "" && a.State != "unknown" inside the on-head check — and change the #14821 fixture to ci("unknown"), review("unknown") so it fails without the fix. All the cases you intend still hold: #16524 (step + changes_requested), "decision with review findings open", "CI red, review approved" are all real non-approved / non-passing states.
2. Go caption always says "a maintainer"; the JS word names the author (low)
src/pkg/dashboard/stage_tracker_caption.go:846 returns "Waiting for a maintainer · <step> on GitHub" unconditionally, while tracker-words-core.js stageTrackerUnauthorizedWho writes @<author> for a human author. The PR body says the word reads "Waiting for @clubanderson · rerun shared ci on GitHub" — that is true of the SPA's rendered chevron but not of the JSON caption field, so any consumer of caption (API clients, tooltips that use it) reads a different sentence than the chevron. If that split is intentional, a one-line note at :846 would save the next reader the same check; otherwise stageTrackerCaption already receives author and could name them when the kind is human.
No security-relevant surface touched; changelog fragment and webassets copy are in place. I have not changed anything on the branch.
— hive: agent=reviewer backend=copilot model=claude-fable-5.1 copilot=1.0.88
|
Thank you for your contribution! Your PR has been merged. We'd love to hear how your experience was: share feedback |
Fixes #16592
What
The Activity Monitor drew hive's own #16524 ("Awaiting human: rerun shared ci",
hive:unauthorized, CI unknown, review findings open) with its lit chevron in the Merge queue column. Nothing about that PR is queued, merging or merge-ready: a maintainer has to rerun shared CI on GitHub. The chevron now lights PR in progress — "Waiting for @clubanderson · rerun shared ci on GitHub", person icon kept — and the Merge queue column is empty.Rule
Merge queue is only for rows actually queued / merging / merged (merge_gate, merge_train, soak), or a merge-ready PR (review approved, CI green on this head) parked on a merge decision. Any 085 PR whose ask is a person's move on the PR (an
unauthorized:<step>reason, or review not approved / CI not green on the current head) renders in PR in progress. A bare 085 with no track evidence keeps the gate, as #14821 decided.How
stageTrackerItemColumn(Go) andstageTrackerAnchor(JS) consult a newstageTrackerPersonActsOnPRpredicate, mirrored on both sides.StageTrackerLinkedPRand both JS nested-PR projections carryunauthorized_stepso the nested chevron reads the same word.Tests
TestStageTrackerPersonActsOnPRColumn: the 🌱 chore: v5 → v6 top-up (2026-11-03) #16524 shape and its twins (merge-ready, queued, bare decision, stale head evidence, issue, 070).TestStageTrackerRowsfixture: 🌱 chore: v5 → v6 top-up (2026-11-03) #16524 rendersactive detouredin PR in progress, Merge queuetodowith no words, one lit chevron, person icon kept; merge-ready / train twins keep the gate; nested row carries the word.make dashboard-gen-check,go test ./pkg/dashboard/ -run StageTracker,gofmt -lclean on touched files.Note on the Review chevron
#16524's review verdict really is
changes_requested(aggregate verdict from the verified bundle; the security perspective's finding is the blocker), so "5/5 · changes requested" is the honest word there — no #16578 sign-off caption applies.