Repository navigation
π delegate: an offline author's PR is not "picked up" by someone else's activity - #18815
Conversation
β¦'s activity "Let hive decide" returned projectbluefin/bluefin-bling#128 to an author whose clanker had been offline for a day (rule return-to-contributor, "the author's clanker has picked the rework up"). delegatedReworkPickedUp counted the tracker's "acknowledged" rework state, which only says the PR's updated_at moved after the request: a maintainer's review on 11-11 moved it. The direction timed out after 30 minutes, and because stageTrackerWaitKind skips every row with a decision record, the PR never reached Waiting on you; it sat five days until closed by hand. - delegatedReworkPickedUp counts only in_progress (the author's relay holds the fix task) and pushed (a new head). An offline author now falls through to the adoption rung. - stageTrackerWaitKind skips a decided row only while its effect has not timed out; a timed-out decision is a person's wait again by the row's real stage. Signed-off-by: Danathar <6772335+Danathar@users.noreply.github.com> Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01BQYnPGFJx81c8JXR3o1sg1
|
[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
|
|
@Danathar this PR is back in your court ( Failing checks:
Reply here or push a new head to this branch (same sign-off); CI and the review swarm re-run automatically and the PR returns to the normal path. Queued in your relay fix queue (2 waiting): fixes are handed out before new work as soon as you have capacity. |
|
CI note: the only non-green check is π Hive Agent: β hive: agent=scanner backend=claude model=claude-sonnet-5-5 claude=2.1.326 |
The Mutation (advisory) job on 291d6d1 was cancelled at its 20-minute timeout while actions/checkout's `git fetch` hung; no code ran. No source change β this empty commit gives the PR a fresh head for CI. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Signed-off-by: Danathar <6772335+Danathar@users.noreply.github.com>
|
Hive status β updated in place; edits do not notify. Full history is in the Activity Monitor. Evidence bundle `#188158eb2c5b` sha256:6526109b0c44510d0853eb6f40f1b9bb5c416055e097180596dc05a5df85f440 β download from the dashboard or `hivectl review evidence #18815`.merged by Hive trusted-author auto-merge: author |
Summary
"Let hive decide" sent a PR back to an author whose clanker had been offline for a day. It mistook a maintainer's review for the author picking the rework up. When nothing happened, the PR also dropped out of Waiting on you for good, so nobody was told.
This PR fixes both:
What happened
Seen on projectbluefin/bluefin-bling#128, on hive build
3b50113. Times are UTC.return-to-contributor. Its comment on the PR says "the author's clanker has picked the rework up".Cause
1. The pickup signal counted anyone's activity.
ContributorOnline || ReworkPickedUp.ContributorOnlinewas false: the author had no activity on the PR within the 15-minute presence window.ReworkPickedUpcomes fromdelegatedReworkPickedUp, and it counted the rework tracker'sacknowledgedstate.updated_atmoved after the request (stage_tracker_rework.go:236-250, "somebody replied or edited"). The maintainer's review on 11-11 moved it.updated_atcannot say who acted.2. A timed-out decision left Waiting on you for good.
stageTrackerWaitKindskips every row that has a decision record ("a decided row is the hive's to act on next").stage_tracker_decided.go:206-208). So the row was nobody's to act on, and the badge, the pill and the MCP'swaiting_on_youall left it out.What changed
delegatedReworkPickedUpcounts only:in_progress: the author's relay holds the fix task;pushed: a new head.With an offline author and neither of those, the ladder falls through to rung 5 (adopt in supersede mode), as Let hive decide adopts PRs whose issue was already fixed and merged (#15219 / #15240)Β #15289 designed.
stageTrackerWaitKindstill skips a row with a pending or landed decision, but not one whose effect timed out. That row's wait kind comes from its real stage again (085 β needs-decision, 090 β ready-for-human-merge, and so on).The
Input.ReworkPickedUpcomment and a changelog fragment are updated to match.The tracker's own
acknowledgedchip on 071/072 rows is unchanged. It is a display hint ("something moved"), not a claim about who acted.Not in this PR
abandoned_prsrecord) runs only for 072 PRs (adoptContributorPR). Add in-progress issue pills and restart reset buttonΒ #128 was at the owner-designed sign-off-attribution 085.decidePROwnershipGatedreports any standing review decision as a person's review. That's separate from this PR and from π automerge: an outdated change request waits on its reviewer, not the authorΒ #18810, which handles a change request left on an older commit.Tests
delegated_rework_pickup_test.goreplays Add in-progress issue pills and restart reset buttonΒ #128. Its first assertion checks that the tracker really reads the maintainer's review asacknowledged. It then checks:acknowledgedstate as a pickup;stage_tracker_waiting_test.go: a timed-out decided row at 085 counts as needs-decision, and a decided row with no timeout still doesn't count.stage_tracker_decided_test.go: the existing "timed out falls back to the real stage" case now also checks the row is a ready-for-human-merge wait.Only
gofmt -lwas run locally (AGENTS.md bans localgo test/go build/go vet); this PR's CI is the test run.π€ Generated with Claude Code
https://claude.ai/code/session_01BQYnPGFJx81c8JXR3o1sg1