Repository navigation
🐛 fix(dashboard): evidence-based remediation state on the PR in progress chevron - #16527
Conversation
…ess chevron
A red PR (changes requested / CI failing) read nothing in the PR in
progress column: waiting_fixer lit CI / Review, and the persistent push
marker overrode it with 'New head pushed' even when the red review landed
after that push. The chevron now carries Changes requested · unassigned
(bright, still), Remediating · <actor> · <age> only on pickup evidence
(fix dispatch, working agent on the PR, claim ledger, contributor relay
claim) and a done-style Fix pushed <sha> once a head lands after the red
signal. The item exposes remediation {actor, kind, since, source}.
Fixes #16525
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
|
|
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. Reviewed by Reviewer notesRead the new What I verified:
CI was still in progress when I read this; nothing above depends on it. No findings. |
There was a problem hiding this comment.
Read the new file and its callers at head 9367cea; this looks correct to me. What I checked:
- Wiring (
stage_tracker_build.go:548-551):activityItemis a copy of the row made afterPush/Rework/Claimantare set (:209,:232,:478), and the new line copies the freshly rebuiltCIActivity/ReviewActivityonto it beforestageTrackerApplyRemediationruns, sostageTrackerRedSincereads the current-head verdicts, not stale ones.activityItem.Repois already qualified (:525), matching thequalify(action.Repo)/qualify(fix.Repo)comparisons inremediationFromAgentNowandremediationFromFixDispatch. - State strings match their producers:
"changes_requested"atstage_tracker_activity.go:454,464;"failing"at:147;pushStatePushed/reworkStatePushed/reworkStateInProgressconstants exist. - "Push predates the red review is not a fix" — the headline claim — rests on
StageTrackerPush.Sincebeing the landing time (stage_tracker_push.go:45, "sinceis when it landed"), soremediationFromPush'sp.Since.After(redSince)is the right comparison.remediationRedPR(push −3h, review −40m) +TestStageTrackerRemediationUnassignedWithoutEvidencecover exactly that path, andTestStageTrackerRemediationEvidencePushcovers the positive one. go build ./pkg/dashboard/passes at head; gofmt clean. Body matches the diff (no claimed behaviour missing).
One non-blocking note: remediationKindComment is declared and appears in the what map but no derivation emits it — the comment on the constant says so, so this is intentional reservation, not dead-code drift.
— hive: agent=reviewer backend=copilot model=claude-fable-5.1 copilot=1.0.88
|
merged by Hive trusted-author auto-merge: author @clubanderson holds role owner and GitHub permission admin on repo; CI green at 9367cea. |
|
Thank you for your contribution! Your PR has been merged. We'd love to hear how your experience was: share feedback |
Fixes #16525
Problem
Six 070 PRs with
Review 5/5 · changes requestedshowed an empty PR in progress chevron.waiting_fixerlit the CI / Review column, and the persistentrecorded-pushmarker overrode the code activity with "New head pushed · CI and re-review" even when the changes-requested review landed after that push — the pushed head was the reviewed head, not a fix. The payload said nothing about whether anybody had picked the PR up.Change
New
stage_tracker_remediation.go, applied afterstageTrackerApplyCodeActivityon every PR row:stageTrackerRedSince— when the current head's latest red signal began (review changes requested / CI failing; a 07x row with no verdict on its head is red since it entered the stage).stageTrackerRemediationEvidence— the one place pickup evidence is read, one helper per source, most recent wins:dispatch.Fixesfor this head within TTL (fix-dispatch); a working hive agent whose current non-review action is on the PR since the red signal (agent-now); a hive-agent claim-ledger / kick-history claim since then.rework.in_progress); a clanker claim-ledger / claim-comment claim since the signal.recorded-push); 072 author head after the ask (rework-commit).updated_atbump is not evidence.code_activitystates and chevron words:unassigned→Changes requested · unassigned · next: scanner/CI failing · unassigned · next: @alice/… · fixer queued, 3 ahead · next: @bob's clanker— DisplayColumn 030, bright, still; tooltip: what triggers pickup.remediating→Remediating · scanner · 4m(· waiting for CIwhen the agent is holding for CI) — DisplayColumn 030, pulsing; tooltip: evidence kind, source, since.fix_pushed→Fix pushed f269349 · re-review pending/· CI pending— CI / Review stays current, the 030 step renders done-style with its word.Waiting for @erin · changes requested(unchanged placement).fixing), same-head CI retry and a 07x push with no verdict yet keep their native words.remediation: {actor, kind, since, source, head}(omitempty).tracker-row-strip.js):data-remediationon the 030 cell, pulse only forremediating, done-styleham-fix-pushedcell with caption,code.detail+ evidence in the tooltip. CSS for the two new looks.Tests
TestStageTrackerRemediation*— unassigned (no evidence), CI-only word + person next, kick from fix dispatch (+ TTL expiry), kick from working agent (review action and pre-signal action rejected), claim (relay, claim comment, stale claim rejected, queued → waiting word), push (recorded push →CI pending/re-review pending, 072 rework push), owner wait / observed work / green PR untouched.stage_tracker_remediation_chevron.test.js— the three chevron shapes (unassigned / remediating / fix pushed) + stale head.go test ./pkg/dashboard/green locally;make dashboard-gen-checkclean.Related pipeline stall (why nothing is remediating these PRs right now): #16526.