Repository navigation
🐛 Activity Monitor: show Let hive decide beside the lead; action pills read as buttons - #16544
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
|
|
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 Intent alignment (reviewed head The diff does what #16543 asks: lead + "Let hive decide" inline on a human-decision row, filled
Review findings from Correctness review (read at head The two-pill render and click dispatch check out: both buttons carry
(and regenerate |
…ion pills read as buttons A row parked on a human decision rendered only the server's lead pill (Re-author); Let hive decide sat under ⋯ → Decide…, invisible. Both now render inline, lead first, via stageTrackerLeadCompanion / stageTrackerLeadActionsOnRow, wired to the same data-ham-act dispatch. Inline action pills were the same muted outline as record pills (adopted from #N, base branch) and read as history. They are now filled buttons — amber primary, neutral secondary, red danger — with a trailing ▸ glyph, hover/focus ring, and a consequence tooltip (server consequence, else a client clause per verb). Fixes #16543 Signed-off-by: Andrew Anderson <andy@clubanderson.com>
…e ratchet) Signed-off-by: Andrew Anderson <andy@clubanderson.com>
e104cf5 to
5f2c089
Compare
There was a problem hiding this comment.
Read tracker-actions.js on e104cf5 end to end for the pill path (render → click → pending/busy). The companion wiring is sound: stageTrackerLeadPillHtml emits data-ham-act="decide:delegate" and the row click handler (tracker-actions.js:896-906) dispatches by that id into stageTrackerDecisionRun, the same path the ⋯ entry uses, and the hover prefetch at :884 matches it too. stageTrackerLeadCompanion derives enable/why from the real decide submenu entry rather than a second source of truth. Looks correct to me, with one verified gap:
Finding (medium, visual only): the busy state never reaches the companion pill.
tracker-actions.js:2360—stageTrackerLeadBusySync(n.querySelector(':scope > .ham-row-head .ham-lead-act[data-ham-act]'), …)passes only the first.ham-lead-actin the row head.tracker-actions.js:2331—on = busyAct && lead.getAttribute('data-ham-act') === busyAct.tracker-actions.js:1889— the delegate preview tags its pend withbusy='decide:delegate'.
On exactly the row this PR targets (lead = decide:re-author, companion = decide:delegate), the first pill's id is decide:re-author ≠ decide:delegate, so on is false and the "Let hive decide" button never gets aria-busy / the Asking the hive… spinner that stageTrackerDelegatePreview's own comment (:1881-1886) and the .ham-lead-act[aria-busy="true"] CSS rule promise. Before this PR the delegate pill was always the only/first pill, so the selector was sufficient. Double-submission is still prevented by _stageTrackerDelegatePreviewing (:1876-1880), so this is feedback, not state. Fix is one line: select .ham-lead-act[data-ham-act="<busy>"] (or querySelectorAll and sync each).
The new tests in stage_tracker_lead_action_test.go cover render/ordering/tone but not the busy sync, which is why CI is green on it.
— hive: agent=reviewer-2 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 #16543
What
stageTrackerLeadCompanionadds the Decide… submenu'sdecide:delegate("Let hive decide") beside the server lead when the row is parked on a human decision (decision_neededon the item / its one parked PR,stage_actor.kind == "human", or stage 085) and the lead isn't already delegate.stageTrackerLeadActionsOnRowreturns[lead, delegate];stageTrackerLeadActionHtmlrenders both, lead first. Rows not parked on a human decision keep a single pill; passive chips andstageTrackerLeadAction/stageTrackerLeadActionOnRowkeep their single-action API (tracker-human-attention.js). Both buttons carrydata-ham-actand run the existing delegated click path (stageTrackerDecisionRun), same as the ⋯ entries..ham-lead-actis now a filled chip (ham-act-primaryamber,ham-act-secondaryneutral for delegate,ham-menu-dangerred for a server-led close) with a trailing ▸::afterglyph, pointer cursor, hover brightness + focus ring; disabled stays dashed/faint and drops the glyph. Record pills (.ham-badge: adopted from, base branch, …) are untouched muted outlines.titleis the serverconsequence; when absent,STAGE_TRACKER_ACTION_CONSEQUENCESsupplies a clause per verb (re-author, delegate, adopt, accept-triage). Delegate's hover no longer says just "delegate".Tests
stage_tracker_lead_action_test.go: two buttons for a lead+delegate row (order, classes, target, enabled state, reader-disabled state), one for a parked issue whose lead is delegate, Re-author fixture (✨ feat(github): enforce pin-only exception in merge-request relay #16456 shape) with consequence tooltip fallback and server override, red danger lead, record class never on actions; fixture-wide maintainer-rows test asserts 2 only when a Decide… target exists.stage_tracker_decided_ui_test.go: the adopted-from record chip carriesham-badgeand neverham-lead-act/<button>.make dashboard-gen-checkclean;go test ./pkg/dashboard/ -run StageTrackergreen; gofmt clean on changed Go files.