Repository navigation
🐛 fix(tracker): PRs at CI / Review ride no merge train; overlap hint instead - #15658
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 |
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. |
|
stage 080-awaiting-merge-admission → 071-needs-author: optional-check-failing:test (dashboard 2/4) |
|
stage 080-awaiting-merge-admission → 071-needs-author: label:needs-rebase |
ad1193e to
a796fef
Compare
…iders The merge train is a merge-queue concept: only PRs at 080+ ride it. A PR still at CI / Review whose files overlap a queued PR gets a lightweight "overlaps #N — will queue behind it" hint instead of a rail or ordinal. The header counts only the riders in this payload (never the planner's wider grouping — hive's own page read "2 of 9 PRs here · not in this view: 7 more" over two queued PRs), says where undrawn members are (on another page / hidden by the filter / not in this payload), and its hover explains the order: shared files, merge one at a time, each rebased onto the base only after the PR ahead merges; a member reads "Nth of M · after Fixes #15653 Signed-off-by: Andrew Anderson <andy@clubanderson.com>
a796fef to
1293fa2
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. |
|
stage 080-awaiting-merge-admission → 071-needs-author: conflicting |
|
Thank you for your contribution! Your PR has been merged. We'd love to hear how your experience was: share feedback |
|
stage 080-awaiting-merge-admission → 050-review-pending: required-check-pending:coverage |
Fixes #15653
Why the train appeared
hive's own page (roll 9 = v5, 2026-10-29 ~20:00 EDT) led
⛁ merge train · 2 PRs · 1st: #15637 — overlapping: sidebar.html, index.htmlunder #15637 and #15633 while both chevrons read CI / Review.merge-train:queued-behind#15637:2/2.stageTrackerMergeTrainPostPass) dropped only PRs under repair (🐛 fix: require verified signature for email-only hive commit identity match #15405 / 🐛 tracker: PRs under repair leave the merge train #15616) and seated 050/060 members as provisional slots (🐛 tracker: chevron and caption agree — queue slot is a suffix until the PR is ready #15581), still drawing the rail, ordinal and header.Change
The merge train is a merge-queue concept: only PRs at ≥080 ride one, and the header counts only them.
stageTrackerBelowMergeQueue: a planner-ordered PR below 080 leaves the train in the view; the 080+ members close ranks (size, position, front, ahead, members, and the front's gate/caption recomputed); one member left is no train.overlap: {behind, with, files, note}instead ofmerge_train(mirrored ontolinked_prs[]): the SPA draws one muted badge overlaps 📖 community: list the Twitch live stream everywhere community links live #15637 with hover Not in the merge queue yet: overlaps 📖 community: list the Twitch live stream everywhere community links live #15637 (sidebar.html, index.html) — will queue behind it (the planner's front: … merges ahead of it once both are ready). No rail, ordinal or "1st:" header.merge_train.provisionaland "merges first once it is ready" removed from the server; the SPA keeps the provisional words only as a fallback for a stale payload.dashboard-sections.md§ Merge train,api-reference/dashboard-misc.md) and changelog fragment.Not changed: the planner still orders CI-pending PRs with ready ones on the lane side (a ready 080 PR can wait behind a CI-pending front); noted in the issue as a separate automerge question.
Tests
TestStageTrackerMergeTrainCIReviewPairRidesNoTrain— 📖 community: list the Twitch live stream everywhere community links live #15637/🌱 refactor: split stage tracker UI tests #15633 exactly: nomerge_train, noleft_train, hint on both rows, payload shape.TestStageTrackerMergeTrainCountsOnlyMergeQueueMembers— 080/050/080: train of 2, "1st of 2", the 050 row hints at the front.TestStageTrackerMergeTrainPendingFrontLeavesLoneRider— 050 front + 080 sibling: no train, 080 row back to its own caption.stage_tracker_overlap_hint.test.js— badge, hover, tooltip line, precedence over train / left-train.go test ./pkg/dashboard/...andnode --test src/pkg/dashboard/*.test.jsgreen; bundle regenerated fromstatic-src/.Follow-up: "what does 'not in this view' mean?"
The header
merge train · 2 of 9 PRs here · 1st: #15516 · not in this view: 7 morecounted the planner's whole overlap group (9) minus the rows drawn. The seven were PRs the planner grouped by shared files but that never carried a train record in the payload — still at CI / Review (below 080), under repair, or no longer listed — so they were counted but never seated. Yes: members share files, so each can only rebase onto the base after the one ahead merges.Now:
size/memberscount only the PRs riding in this payload (≥080). If the page does not draw every member the header says why:on another page: #N,hidden by the filter: #N,not in this payload: #N. The header's hover states the mechanics (shared files → merge one at a time, in order; each rebased only after the PR ahead merges); a member's hover showsPosition: 3rd of 4 · after #X merges.Tests: Go
TestStageTrackerMergeTrainSizeCountsRidersInPayload; JS header cases (all on page / other page / filtered / payload-missing / no context) instage_tracker_merge_train_test.go.