Repository navigation
🌱 dashboard: defer merge-gate pills until the PR reaches the gate - #14651
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 |
|
@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. |
Open PR overlap checkBase branch: Conflicts are checked with every PR merged onto the current tip of
|
|
@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. |
There was a problem hiding this comment.
Read the diff against v5 and the consumers it touches; ./pkg/dashboard builds at 22f09db. Looks correct to me.
What I verified:
- Server split:
stageTrackerPillsDeferred(stage_tracker.go:677-682) gates only PRs and mirrored issues onstage < AwaitingMergeAdmission; own-stage issues keep every pill, as the body says.it.Mirroredis computed on the line before it is passed in (:1247-1248), so the mirrored case reads the right flag. - The two post-hoc
DeleteFuncstrips for 045/085 and 015 (:1252-1260) still operate onModifiers; at 085 nothing is deferred (≥080), so the needs-decision chip is removed by the chevron rule rather than deferral — no double handling. UnauthorizedStepis still populated when the marker is deferred (:1265), so the tooltip's "Pending at merge gate" line and the gate cell text keep their reason.- The
approve-workflow-runscarve-out (:690-694) matchesprstage.StageReasonUnauthorizedPrefix+UnauthorizedStepApproveWorkflowRuns(reasons.go:52,55) afterHumanApprovedReasonstrips the approver suffix. - Client:
stageTrackerNeedsAttention(index.html:14974-14978) readsmodifiersonly, so deferred pills drop out of the count without a client change; sentinel chip and tooltip both consultdeferred_modifiers(:14243,:14229-14231); nestedlinked_prsand the single-covering-PR projection carrydeferred_modifiersthrough (:14324,:14503,:15754). stageTrackerResolveLinkedPRscopiesDeferredModifiersalongsideModifiers(:1778-1780).
No security, style or docs concerns; dashboard-sections.md and the changelog fragment are updated with the change.
— hive: agent=reviewer backend=copilot model=claude-fable-5.1 copilot=1.0.88
There was a problem hiding this comment.
Conclusion: the server-side split and the SPA wiring implement the owner's rule as the body states it, and I found no behavioural defect — but the PR's own diff breaks TestStageTrackerSentinelLaneWiring, which is what is turning test, test (dashboard 2/4), test (dashboard shuffle 2/3) and coverage red on this head (run 40556454861).
What I read at 22f09db: stage_tracker.go (stageTrackerClassifyModifiers / stageTrackerModifiers 700–778, stageTrackerPillsDeferred 678, sentinel deferral 1367–1372, stageTrackerResolveLinkedPRs 1778), static/index.html (stageTrackerNeedsAttention 14974, tooltip 14226–14231, stageTrackerSentinelChipHtml 14243, sub-row/via_pr 14321/14503, stageTrackerSentinelPreface 16237), stagelabels.go:275 (AwaitingMergeAdmission = 80), prstage/reasons.go:52. go build ./pkg/dashboard/ ./pkg/github/ passes locally.
Finding (high — CI red from this diff): src/pkg/dashboard/stage_tracker_sentinel_lane_test.go:267 still asserts index.html contains the literal
if (item.sentinel_alert) lines.push(stageTrackerSentinelTip(item));
but this PR rewrites that line at static/index.html:14231 to if (item.sentinel_alert && deferred.indexOf(STAGE_TRACKER_SENTINEL_CHIP) < 0) lines.push(...). CI log: stage_tracker_sentinel_lane_test.go:271: index.html missing sentinel lane wiring "if (item.sentinel_alert) lines.push(stageTrackerSentinelTip(item));" → --- FAIL: TestStageTrackerSentinelLaneWiring. Update the expected string in the wiring test to the new guard (the PR already touched this file, so it is a one-line follow-up). The coverage failure is downstream of the test failure.
Verified correct, for the record:
stageTrackerNeedsAttention(index.html:14974) readsitem.modifiersonly, so deferred merge-gate pills genuinely drop out of the "Needs attention" count and filter — matches the body.- The sentinel chip is suppressed only when
deferred_modifierscarriessentinel(index.html:14243), whilestageTrackerSentinelPreface(16237) still keys offsentinel_alert, so the Decide… dialog preface keeps the alert as the Go comment atstage_tracker.go:1367intends. stageTrackerUnauthorizedis still computed whenunauthorizedsits in either list (stage_tracker.go:1265), so the tooltip's "Unauthorized: …" line survives deferral.- Mirrored issues (050+) follow the PR via
stageTrackerPillsDeferred(kind, n, mirrored); issues on own stages (045 etc.) keep every pill — as the body's table says.
Not blocking: the other failures in the same job (api_repo_item_decision_test.go:106, health_disk_test.go:187) do not touch files in this diff; I did not trace them.
— hive: agent=reviewer-2 backend=copilot model=claude-fable-5.1 copilot=1.0.88
|
hive/ci-report @ 22f09db — The PR rewrites that exact-string-guarded line in if (item.sentinel_alert && deferred.indexOf(STAGE_TRACKER_SENTINEL_CHIP) < 0) lines.push(stageTrackerSentinelTip(item));but the guard at The hive does not push to author branches, so this is a report only. 🐝 Hive Agent: — hive: agent=scanner backend=copilot model=claude-fable-5.1 copilot=1.0.88 |
22f09db to
04a5294
Compare
|
@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. |
There was a problem hiding this comment.
Read the Go diff against the v5 tree and the branch builds. The classification itself looks correct to me; the work here is the rebase.
Rebase facts (base ee64433, 30 behind): git merge-tree v5 fix/pills-until-gate now conflicts in three files — src/pkg/dashboard/stage_tracker.go, static/index.html and src/docs/dashboard-sections.md. The v5 commits that moved the same regions are #14689 (deee564ac, attention-first ordering), #14669 (cc4cc1680), #14699 (f6988e43c) and #14657 (27e655124).
Good news on #14689: its stageTrackerNeedsAttention / stageTrackerAttentionTier read it.Modifiers only and the comment already says "Only the shown Modifiers count, so a merge-gate pill …" — so once this PR's split lands, deferred pills drop out of the attention count and tier with no extra change. stageTrackerHumanMods there still lists modNeedsHuman/modNeedsDecision, which is right for issues on their own stages (never deferred, per stageTrackerPillsDeferred).
What I verified in the diff:
stageTrackerPillsDeferredkeys offstagelabels.AwaitingMergeAdmission(number 80,stagelabels.go:275), so 080/085/090/095/100 show their gate pills and 085-parked PRs keep "needs-decision" visible.it.Mirroredis assigned before thestageTrackerModifierscall infinish, so mirrored issues are classified with the right flag.stageTrackerUnauthorizedis still computed whenunauthorizedis deferred (slices.Contains(it.DeferredModifiers, …)), so the gate word / Decide… preface keep their text.modSentinelis prepended only toDeferredModifiers, neverModifiers, matching the comment that the chip is drawn fromSentinelAlert.
Also overlaps #14573 (same author): that PR adds a prowWIPLabel case and a hold exclusion inside the old stageTrackerModifiers; whichever lands second re-applies it inside stageTrackerClassifyModifiers.
— hive: agent=reviewer backend=copilot model=claude-fable-5.1 copilot=1.0.88
04a5294 to
8b10383
Compare
Split tracker pills into actionable-now (hold, draft, stalled, regressed, unauthorized on held workflow runs) and merge-gate (sentinel, needs-human, needs-decision, merge-time unauthorized). Merge-gate pills are held back in deferred_modifiers while the PR is before stage/080 and neither render nor count under Needs attention; they render as before once the PR reaches the gate. Labels are untouched. Fixes #14650 Signed-off-by: Andy Anderson <andy@clubanderson.com>
…ooltip line Signed-off-by: Andy Anderson <andy@clubanderson.com>
…S harness Signed-off-by: Andy Anderson <andy@clubanderson.com>
8b10383 to
7b74e48
Compare
|
@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. |
|
merged by Hive trusted-author auto-merge: author @clubanderson holds role owner and GitHub permission admin on repo; CI green at 7b74e48. |
|
Thank you for your contribution! Your PR has been merged. We'd love to hear how your experience was: share feedback |
Fixes #14650
Owner's rule (2026-10-27)
What changes
Pills in the far-right column of the Hive activity monitor are now split into two classes, computed server-side in
stageTrackerModifiers(eachstageTrackerModifiercarries an explicitActionableNow). The payload carries drawn pills inmodifiersand held-back ones indeferred_modifiers. A deferred pill is absent — no chip, no breathing chevron, no Needs attention count — until the PR's stage reaches the merge-gate band (stagelabels.AwaitingMergeAdmission, stage/080+, the same boundary adoption yields at); the row tooltip alone says "Pending at merge gate: …". The merge-gate cell text ("Sentinel alert · needs human", "Awaiting human: …") is unchanged. Issues on their own stages keep every pill; an issue riding its covering PR follows the PR. Labels are untouched.holddraftstalledregressedstage ↓unless parked on a person, as today)needs-rebase,blocked,triage needed,possible dupeunauthorized—approve-workflow-runsunauthorized— merge / branch-update / unknown stepsentinel(notice and realsentinel-alert)sentinel_alertstays truthful in the payloadneeds human(needs-humanlabel)needs human decision(needs-decisionpreviews)Notes
bands.gocounters read labels, not tracker modifiers, so they are untouched; the tracker's own Needs attention button follows the rule because it counts drawn modifiers only.·dot affordance added: no such single-glyph widget existed in the pill column, so per the brief none was invented; the tooltip line is the only hint.src/docs/dashboard-sections.md; changelogchangelog.d/changed-pills-until-merge-gate.md.Tests
stage_tracker_pills_gate_test.go: table-drivenTestStageTrackerModifiers_DeferredUntilMergeGate(sentinel/needs-human/unauthorized across 050 / 072 / 080 / 085 for PRs, mirrored and own-stage issues),TestStageTrackerClassifyModifiers,TestStageTrackerDeferredPillsPayload(end-to-end payload),TestStageTrackerDeferredPillsJS(Node fixture assertions overstageTrackerChipsHtml,stageTrackerSentinelChipHtml,stageTrackerTooltip,stageTrackerNeedsAttention,stageTrackerMotion). Notime.Sleep.