Skip to content

πŸ› fix(stage): word a hive hold's release step for the gate that set it - #16821

Merged
hivecommons-hive[bot] merged 1 commit into
hivecommons:v6from
Danathar:fix/hold-step-by-source
Nov 5, 2026
Merged

hivecommons-hive[bot] merged 1 commit into
hivecommons:v6from
Danathar:fix/hold-step-by-source

Conversation

@Danathar

@Danathar Danathar commented Nov 5, 2026

Copy link
Copy Markdown
Contributor

Executive summary

When the hive puts hold on 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 … and the lane merges on its next pass … 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 … 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

Hold set by the hive for… Before After
reporter trust "Release … lane merges … until … reviewed the issue's reporter" unchanged
a human-merge path same reporter-trust text "the hive never merges it, so a maintainer reviews it and merges it by hand; removing hold does not hand it back to the lane"
a gate known only from the label event same reporter-trust text ("for its own gate") "its notice on this PR names the gate. Resolve that, then remove the hold label"
popover question (human-merge path) "Release the hold (the lane then merges it…), or keep it?" "…the hive never merges it. Review it and merge it by hand, or close it?"

Why

  • stagelabels.DecisionActions printed one fixed sentence for every hold the hive set itself, written for the reporter-trust gate.
  • automerge.holdProvenance knew only two sources, reporter-trust and label, so a human-merge-path hold read as an unexplained "its own gate".

How

  • stagelabels: new source human-merge-path and HoldProvenance.HumanMergePath(). selfImposedHoldStep picks 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

Test Covers
TestSelfImposedHoldStepMatchesItsGate reporter trust / human-merge path / label-only text
TestHoldProvenance (+3 rows) App notice β†’ human-merge-path; someone else's notice β†’ label; reporter trust wins
TestStageTrackerDecisionHumanMergePathHold popover asks for a hand merge, never promises a lane merge
Mutants 6/6 caught
go test stagelabels, github/..., stageengine/..., writeback, dashboard pass

Notes

  • Dropped from the same report: the duplicate stage-needs-decision comments on finpilot#627 stopped after the 11-05 06:36Z pair (no new pair on head 0149d18).

πŸ€– Generated with Claude Code

https://claude.ai/code/session_01BQYnPGFJx81c8JXR3o1sg1

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 … 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>
@kubestellar-prow kubestellar-prow Bot added the dco-signoff: yes Indicates the PR's author has signed the DCO. label Nov 5, 2026
@kubestellar-prow

Copy link
Copy Markdown
Contributor

[APPROVALNOTIFIER] This PR is NOT APPROVED

This pull-request has been approved by:
Once this PR has been reviewed and has the lgtm label, please assign kellyaa for approval. For more information see the Code Review Process.

The full list of commands accepted by this bot can be found here.

Details Needs approval from an approver in each of these files:

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

@kubestellar-prow kubestellar-prow Bot added the size/L Denotes a PR that changes 100-499 lines, ignoring generated files. label Nov 5, 2026

@hivecommons-hive hivecommons-hive Bot left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

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.HumanMergePathMarker is <!-- 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 just SelfImposed() (automerge_stage_labels.go:1016, reporter_trust_release.go:156), so a hold now attributed to human-merge-path is not released into the lane β€” consistent with the new wording "removing hold does not hand it back to the lane".
  • stagelabels/human.go:922-937 selfImposedHoldStep: 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

@github-actions

github-actions Bot commented Nov 5, 2026 •

Copy link
Copy Markdown
Contributor

Open PR overlap check

Base branch: v6

Conflicts are checked with every PR merged onto the current tip of v6, so a PR that only needs a rebase is not reported as conflicting with its siblings.

PR Title Relation Files
#16828 πŸ› fix(blocked): one Unblocked note per lift, and not "actionable" when un-admitted same files src/docs/labels-and-control-signals/reference.md
#16811 πŸ› fix(workflows): verify "the App cannot push workflows" before parking on it same files src/docs/dashboard-sections/activity-monitor.md
src/docs/labels-and-control-signals/reference.md
#16816 πŸ› fix(adoption): refuse a fixer push that undoes the adopted PR's change same files src/pkg/dashboard/stage_tracker_decision_pr.go

@hivecommons-hive

hivecommons-hive Bot commented Nov 5, 2026 •

Copy link
Copy Markdown
Contributor

Hive status β€” updated in place; edits do not notify. Full history is in the Activity Monitor.

Review findings from reviewer (folded here because this head already carries its hive review):

Looks correct to me.

What I verified (v6 tree, head 21cb358):

  • automerge/reporter_trust_release.go:71-75 β€” the new branch sits after the reporter-trust match and before the label-only fallback, so an App-set hold that the reporter-trust notice does not explain is attributed to the human-merge-path refusal only when the App itself posted <!-- hive-human-merge-path --> (humanMergePathNoticed, :81-88, checks the author login). Not time-bounding that comment is consistent with pkg/github/human_merge_paths.go:20-22: the notice is posted once per PR and the label re-added on later heads. The only caller (automerge_stage_labels.go:1015) consumes the provenance unchanged.
  • stagelabels/hold.go:41 β€” HoldSourceHumanMergePath = "human-merge-path" follows this package's existing hyphenated convention ("reporter-trust", :40). pkg/github/hold_sources.go:13 uses "human_merge_path" for its own HoldSources set β€” a separate namespace, pre-existing divergence, not introduced here.
  • stagelabels/human.go:916-937 β€” selfImposedHoldStep keeps the previous wording byte-for-byte for reporter-trust, and the label-only fallback no longer promises "the lane merges on its next pass" for a hold whose gate it cannot see. That matches the behaviour: DecisionActions can't know the gate, so it stops asserting the outcome.
  • dashboard/stage_tracker_decision_pr.go:82-86 β€” StageTrackerHold.Source exists (stage_tracker.go:896) and decisionHoldReason (:320-327) returns Reason when set, so the popover reads "…because it changes a human-merge path (auto_merge.human_merge_paths): the hive never merges it."
  • Tests cover each new branch (hold_step_source_test.go, reporter_trust_release_test.go, stage_tracker_hold_human_merge_test.go); docs and changelog fragment updated.

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`.

@hivecommons-hive
hivecommons-hive Bot merged commit 869cd2d into hivecommons:v6 Nov 5, 2026
72 of 76 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

dco-signoff: yes Indicates the PR's author has signed the DCO. size/L Denotes a PR that changes 100-499 lines, ignoring generated files.

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant