Repository navigation
π fix(stage): word a hive hold's release step for the gate that set it - #16821
Conversation
projectbluefin/utah#1097 and finpilot#627 carry a hold the human-merge-path refusal set, but the 085 comment and the popover gave the reporter-trust step ("the lane merges on its next pass β¦ until a maintainer has reviewed the issue's reporter"). Hold provenance now records human-merge-path from the App's notice, and the release step is worded per source. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01BQYnPGFJx81c8JXR3o1sg1 Signed-off-by: Danathar <Danathar@users.noreply.github.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 |
There was a problem hiding this comment.
Read at head 21cb358 against v6, with holdProvenance in full and the two release call sites. Looks correct to me.
reporter_trust_release.go:67-74: the human-merge-path branch runs only after the actor is confirmed to be the App bot (line 58-60) and after the fresh reporter-trust notice check (line 64), so a reporter-trust hold with a stale human-merge-path notice on the thread still reads as reporter trust, and a person's hold is untouched.hgithub.HumanMergePathMarkeris<!-- hive-human-merge-path -->(human_merge_paths.go:22), the marker the refusal actually posts.- The auto-release path is already gated on
prov.ReporterTrust(), not justSelfImposed()(automerge_stage_labels.go:1016,reporter_trust_release.go:156), so a hold now attributed tohuman-merge-pathis not released into the lane β consistent with the new wording "removingholddoes not hand it back to the lane". stagelabels/human.go:922-937selfImposedHoldStep: the reporter-trust branch reproduces the previous sentence verbatim, so existing 085 comments for that gate are unchanged, as the body's table says.go build ./pkg/github/automerge/ ./pkg/stagelabels/ ./pkg/dashboard/passes at this head.
Open question (not a defect I verified): stage_tracker_decision_pr.go:82-86 changes only the popover question for a human-merge-path hold to "merge it by hand, or close it?", while the rest of that case (Blocker, BlockerReason = hold:hive, and whatever button set the front end derives from hold:hive) is unchanged β so the popover may still lead with an enabled Release hold button under a question that no longer mentions releasing. If that is intended (releasing is harmless since the lane never merges it anyway), a word in the docs would stop the next reader from wondering; if not, the button set for this source is the follow-up.
β hive: agent=reviewer-2 backend=copilot model=claude-fable-5.1 copilot=1.0.88
Open PR overlap checkBase branch: Conflicts are checked with every PR merged onto the current tip of
|
|
Hive status β updated in place; edits do not notify. Full history is in the Activity Monitor. Review findings from Looks correct to me. What I verified (v6 tree, head
No defects found. I have not approved or merged anything. Evidence bundle `#1682121cb358` sha256:f60c883644a759705eefbb4e1a9aa1830ea422197a4fdaf4798ddd92915b0630 β download from the dashboard or `hivectl review evidence #16821`. |
Executive summary
When the hive puts
holdon a PR because it changes a human-merge path, the 085 comment and the decision popover described it as a reporter-trust hold: "Release it β¦ and the lane merges on its next pass β¦ or keep it until a maintainer has reviewed the issue's reporter". The hive's own notice on the same PR says "Hive will not merge this PR β¦ a person must merge it" (projectbluefin/utah#1097, projectbluefin/finpilot#627).This PR records the human-merge-path refusal as the hold's source, and words the release step for the gate that actually set the hold.
What
holddoes not hand it back to the lane"holdlabel"Why
stagelabels.DecisionActionsprinted one fixed sentence for every hold the hive set itself, written for the reporter-trust gate.automerge.holdProvenanceknew only two sources,reporter-trustandlabel, so a human-merge-path hold read as an unexplained "its own gate".How
stagelabels: new sourcehuman-merge-pathandHoldProvenance.HumanMergePath().selfImposedHoldSteppicks the release step by source.automerge/reporter_trust_release.go: when the App set the hold and no fresh reporter-trust notice explains it, the App's<!-- hive-human-merge-path -->notice on the thread does. The notice is posted once per PR and the label is re-added on later heads, so it is not held to the 2-minute slack.dashboard/stage_tracker_decision_pr.go: the popover question for that source.Tests
TestSelfImposedHoldStepMatchesItsGateTestHoldProvenance(+3 rows)human-merge-path; someone else's notice βlabel; reporter trust winsTestStageTrackerDecisionHumanMergePathHoldgo teststagelabels, github/..., stageengine/..., writeback, dashboardNotes
π€ Generated with Claude Code
https://claude.ai/code/session_01BQYnPGFJx81c8JXR3o1sg1