Repository navigation
🐛 fix: unclaimed proposal / kind/design issues derive stage/018-open-for-discussion - #17220
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
|
There was a problem hiding this comment.
docs-currency (read at head 1143dcf)
The doc surfaces in the diff agree with the code: the stage-labels.md:98 row, the activity-monitor.md:71 caption, governor-thresholds.md:5, ADR-0025 and stage-engine-plan.md all match unclaimedIssueStage (src/pkg/stageengine/derive_issue.go), the transition_table.go edges, the IsRegression change (stagelabels/store.go:1216) and the caption text in stage_tracker_caption.go:825-832. Every doc that #17150 names is updated.
One label-reference gap, advisory (in-scope, low):
src/docs/labels-and-control-signals/reference.mdis the per-label reference and has no entry forproposal/kind/design, which now drive stage derivation (src/pkg/github/issue_stage_labels.go:306discussionLabels). The siblingepicentry (reference.md:515-519) documents exactly this kind of effect ("the epic staysstage/010-actionable"), and thehive/partially-addressedentry (reference.md:561) still says the issue "stays in the actionable pool" — after this PR a labelled proposal lands at 018 instead (derive_issue.goissueProgress). Suggested fix: add a#### \proposal`, `kind/design`entry under *Issue admission and triage* (Kind: display/stage derivation, not a gate; derivesstage/018-open-for-discussionwhen unclaimed; config: fixed list inissue_stage_labels.go), and append "(orstage/018-open-for-discussionfor aproposal/kind/designissue, #17150)" to thehive/partially-addressed` paragraph.
— hive: agent=adjudicator-5 backend=copilot model=claude-fable-5.1 copilot=1.0.88
|
Hive status — updated in place; edits do not notify. Full history is in the Activity Monitor. Review findings from Looks correct to me; nothing in the diff needs changing. The two red checks are infrastructure, and the carried-over "changes requested" question is already answered by #17150's own proposed rule. Read the engine, label-store and dashboard code at Verified against the tree
Red checks are not this diff
Carried-over question from #17171 One follow-through for a maintainer after merge, not a code change: #17054 currently carries 🐝 Hive Agent: Review findings from Read the engine, label and dashboard code at Carried-over items — resolved
Verified on this head
Rebase hazard (not a defect) Nothing here blocks landing from my side. I have not approved, closed or labelled anything. — hive: agent=reviewer-2 backend=copilot model=claude-fable-5.1 copilot=1.0.88 Evidence bundle `#17220d4d4c26` sha256:42163bcb99dce55633822ad18b359b766b4302c8039d2ef363c6e3a3578e970f — download from the dashboard or `hivectl review evidence #17220`. |
|
CI triage (fix agent): both red checks on this run are infrastructure failures, not code:
No lint findings were produced; 🐝 Hive Agent: — hive: agent=scanner backend=copilot model=claude-fable-5.1 copilot=1.0.88 |
1143dcf to
2ae750e
Compare
|
CI triage (fix agent), head
The open 🐝 Hive Agent: — hive: agent=scanner backend=copilot model=claude-fable-5.1 copilot=1.0.88 |
There was a problem hiding this comment.
intent-alignment — delta re-review at head 2ae750e (previous head 1143dcf).
The only PR-side change since 1143dcf is one line in src/pkg/dashboard/stage_tracker_phone_attention_first_test.go:35 (adds STAGE_TRACKER_OPEN_FOR_DISCUSSION to the harness prelude); the remaining 112 files in the compare range are the v5 rebase, not PR content. No scope creep.
Still present (in-scope): Fixes #17150 closes the issue with acceptance criterion 5 unmet.
#17150 AC5: "Both #17054 and #17020, replayed through the engine, land on the new stage rather than 010." The rule keys on labels only — discussionLabels = {"proposal", "kind/design"} at src/pkg/github/issue_stage_labels.go:305, and IssueStageInput (issue_stage_labels.go:167-197) carries no title, so 📝 cannot be matched. The PR's own golden pins the gap: src/pkg/github/testdata/issue_stage_goldens/regressions.json:14 (regression/17054-rfc-epic-without-proposal-stays-010) asserts #17054 as labelled today derives 010-actionable. #17054 is still labelled help wanted, Epic, approved-direction (checked now) — so on merge the issue auto-closes while one of its two motivating cases stays "Actionable".
The label-only rule is the right reading of #17150's Proposed rule ("an open issue labelled proposal (and/or kind/design)"), so the code should stand. Requested change — make the landing complete rather than widen the rule:
- Apply
proposalto #17054 as part of landing this PR (gh issue edit 17054 --repo hivecommons/hive --add-label proposal), and record that step in the PR body under a short "Landing" note. The golden atregressions.json:13(17054-rfc-epic-labelled-proposal-is-018) already proves that label flips it to 018. - If labelling the live issue is not acceptable to the fixer, demote the closing keyword to
Refs #17150so the issue does not close with AC5 unmet.
Also noted (not mine, still present at this head, for the correctness lane): src/pkg/stageengine/derive_issue.go:219-221 → :283-284 — an Epic+proposal issue with a stray claim now returns "open for discussion: proposal" and drops the "umbrella issue (…): claim not staged" reason passed in at :220.
— hive: agent=adjudicator-3 perspective=intent-alignment backend=copilot model=claude-fable-5.1 copilot=1.0.88
|
CI triage (scanner), head
Same outage is hitting #17151 and hivecommons/crux#127 at the same time. Nothing to change on this branch; deferring to the next sweep / infra rerun. 🐝 Hive Agent: — hive: agent=scanner backend=copilot model=claude-fable-5.1 copilot=1.0.88 |
🛑 Fix loop escalated — human attention neededThis PR has failed CI on 3 distinct fix attempts (new commits, still red). The hive has stopped dispatching further automated fixes for it. Failing checks: FOSSA license scan, golangci-lint, overlayfs-exec-guard Also red on the base branch / other open PRs (shared breakage, not this PR): build (linux/amd64, true, false, 60), docker (hive) Raw failure evidence (from check-run annotations): Remove the |
✅ Escalation cleared — back in the automated laneThe shared CI breakage this escalation was raised over (build (linux/amd64, true, false, 60), docker (hive)) has cleared — returning this PR to the automated lane. The |
|
Shared CI incident: the red check(s) on this head are GitHub release-download 504s, tracked in #17224 — deferred until the baseline recovers; no changes needed here. 🐝 Hive Agent: — hive: agent=scanner backend=copilot model=claude-fable-5.1 copilot=1.0.88 |
There was a problem hiding this comment.
Read the engine, label-vocabulary and stage-table code at c878544 against the v5 tree before the body. Looks correct to me; the open intent question from #17171 is now answered by this PR's body, so I am not re-raising it.
What changed since the last swarm-reviewed head (1143dcf) — one commit, c878544, 4 lines:
src/pkg/stageengine/transition_table.go:20-21adds015 → 018and018 → 015, whichunclaimedIssueStage(derive_issue.go:276-282) can produce whenever aproposal/kind/designlabel is added to or removed from an untriaged issue.src/docs/labels-and-control-signals/stage-labels.md:152-153draws the same two edges.transition_table_test.go:75-86requires the map and the diagram to agree edge-for-edge; the doc's 68 edge lines vs. the map's 67 entries is the pre-existing duplicates074 --> s085lines, whichdiagramEdgesFromDocdedups.IsRegression(stagelabels/store.go:1216) already listsOpenForDiscussionas a parking bay, so018 → 015is not booked as a fall — consistent with the edge comment.
Carried over, re-verified at this head: Before() compares Number() (stagelabels.go:429), so appending the enum member while numbering it 18 keeps SettlesTriage false for 018; BucketClaimed/BucketInProgress are not routed through unclaimedIssueStage, so a live claim still stages 020/030; issueDiscussionLabel (issue_stage_labels.go:306) is read-only on labels and the claim gate is untouched.
Intent: the body declares Refs #17150 with the remainder stated (AC5 — #17054 needs a maintainer to apply proposal). That settles the Fixes-vs-Refs question the swarm raised on #17171; I have judged the change against that declared scope.
— hive: agent=reviewer backend=copilot model=claude-fable-5.1 copilot=1.0.88
|
The hive's GitHub App lacks |
|
Needs a decision, not a merge: Clear the cause and the hive re-evaluates on its next pass; the stage moves to |
TestCloseOnMergeLedgerSweepSettledClaimsSkipAPI reloads its ledger via LoadClaimLedger, which retires settled claims settledClaimRetention (30d) after their anchor using time.Now. The fixture's fixed merge date (2026-10-09 12:00 UTC) crossed that bound today, so the reload dropped all three claims and the test went red on v5 and every open branch. Use loadClaimLedgerAt with the test clock so retention is evaluated against the fixture's time. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>
There was a problem hiding this comment.
Read the engine, label, store and dashboard code at d4d4c26 against the v5 tree before the body. The derivation change looks correct to me; the one thing worth a maintainer's ten seconds is a sequencing overlap with #17229.
What I verified
src/pkg/stageengine/derive_issue.go:155-230— likely-done, hold, blocked, needs-human, the 017 prior-art park and the 045 decision park all return beforeunclaimedIssueStage(:276-282) is reached, so 018 only applies to a genuinely unclaimed proposal/design issue.BucketClaimed/BucketInProgressare not routed through it, so a live claim still stages 020/030 — the "not a gate" claim holds.src/pkg/stagelabels/stagelabels.go:448—BeforecomparesNumber(), not enum position, so appendingOpenForDiscussionto the persisted enum with number 18 keepsSettlesTriage(:460) false for 018 and the dashboardtrackerNumberClaimedguard consistent.src/pkg/stagelabels/store.go:1216— leaving 018 is never a regression;040 → 018is a regression unlessissueProgress(derive_issue.go:266-271) seeshive/partially-addressed, matching #17150's AC.src/pkg/stageengine/transition_table.go:18-27— the edges added inc878544cover every 018 transition the deriver can produce (010/015 ↔ 018, 018 → 020/045, 045 → 018, 020/030/040 → 018); nothing in the diagram is unreachable.src/pkg/github/prclaims.go:1249—loadClaimLedgerAtexists inv5with the injectable clock the sweep test now uses, sod4d4c26is a valid fix for the wall-clock expiry.
Carried-over finding from #17171: resolved. The docs-currency gap (no 018 paragraph in ADR-0025) is filled — src/docs/adr/0025-central-stage-engine.md is in this diff. The decision question (Fixes vs Refs #17150) is answered by the body: Refs #17150 with AC5 (#17054 needs the proposal label by hand) stated as what remains. That is the right citation for a deliberately partial close.
Overlap with #17229 (sequencing, not a defect)
The last commit here, d4d4c26, changes src/pkg/github/close_on_merge_ledger_sweep_test.go:180-184 in exactly the way #17229 does — same call to loadClaimLedgerAt(ledger.path, nil, func() time.Time { return now }), same removal of reloaded.SetClock. The code is byte-identical; only the explanatory comment text differs. Because the comment lines differ in the same hunk, whichever PR lands second will conflict on that hunk rather than merge as a no-op (hence the conflicts-with-open-pr label). #17229 is the dedicated fix for the red test (rest 3/4) shard and closes #17227/#17228, so it should land first; this PR then needs d4d4c26 dropped or the hunk rebased onto it. No reason to hold this PR on that beyond the rebase.
I have not changed anything.
— hive: agent=reviewer 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 |
…erge - automerge: the v5 unsearchable-repository tests now expect v6's search shape — the PR's own repository is part of the per-author cached search (#15088), and a sweep inside the cache TTL issues no query. - dashboard: stage/018-open-for-discussion (#17220) is an issue stage, not PR coverage; add it to the v6 CoversIssue table test. Reword the v5 tracker.css phone comment so the ☐ glyph is written once in index.html. - crux: split the Linear credential helpers out of agent/manager_tokens.go and the caveman installer out of agent/manager_launch.go to stay under crux's file-size ratchet after the union of both lines. - docs: api-reference citations resynced (--fix); nous-spek-source links the experiment loop in experiment-lab.md (v6 moved it from strategy-lab). Signed-off-by: Andrew Anderson <andy@clubanderson.com>
Supersedes #17171 (adopted in place, re-homed because a review finding on the adopted head sat unaddressed past the stage dwell; original commits by @Karibhou preserved).
Refs #17150 (AC5: #17054 still needs the
proposallabel applied by a maintainer; issue stays open until then)This branch is #17171's head
1143dcf25293as pushed by @Karibhou (forkKaribhou/hive), unchanged. Last lane reason on the original:contributor-lane:changes-requested.Hive fix agent: rebase this branch on
v5, fix the failing checks / review asks, and push to THIS branch. Keep the original author's commits and theirSigned-off-by; add your own sign-off on your commits. Do not open another PR.Open review finding carried over from #17171: the review swarm's latest verdict on
1143dcf25293requested changes (changes_requested); read the swarm's review on #17171, address it here, and let the swarm re-review this PR. Green CI alone is not done.