Repository navigation
π fix(stage): bot Dependency Dashboards and the advisory report get no stage label - #16833
Conversation
mergeraptor's Dependency Dashboards (projectbluefin/gutenprint-printer-app#45, ps-printer-app#50, hplip-printer-app#42) wore stage/045-needs-decision because a triage bot added needs-human. A standing meta issue (bot Dependency Dashboard, Hive Advisory Report) is not work: the engine derives no stage for its bucket, and the out-of-cycle sink classifies it from the fresh read, so the 045 comes off and a webhook cannot re-park it between sweeps. 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 |
Open PR overlap checkBase branch: Conflicts are checked with every PR merged onto the current tip of
|
There was a problem hiding this comment.
Read the diff against the v6 tree at 91ba1dde5 and the callers. This looks correct to me.
What I verified:
- Sweep path agrees with the new constants.
client.go:2018-2026files these issues with the string buckets"hive_advisory"/"dependency_dashboard";issue_stage_labels.go:339passesrow.Bucketthrough, andissueBucketFrom(issue_stage_facts.go:72-77) is a plainIssueBucket(bucket)cast β so the sweep row lands onBucketDependencyDashboard/BucketHiveAdvisorywith no further wiring. Before this PR those strings were carried as an unknown bucket and fell through to thePark.NeedsHumanbranch, which is exactly the 045 described. - Ordering. The new check in
issueStage(derive_issue.go:146) sits ahead ofLikelyDone, hold, blocked and park, matching the sweep, whichcontinues on standing meta beforeisHeld(client.go:2029-2035). No disagreement between the two. - Out-of-cycle path.
stage_facts_sink.go:166reusesstandingMetaIssueKindForwith the same title/author/labels the sweep uses, so a webhook fact classifies identically. go build ./pkg/stageengine/ ./pkg/github/is clean at head40d6e3b.
One non-blocking note: TestStandingMetaIssueDerivesNoStage asserts the bot's needs-human survives the write β good, since that label is not the hive's β but nothing asserts the behaviour for an issue that is both a dependency dashboard and wears a hold label. Reading the code it derives None either way, so I don't think a test is required; mentioning it only because standing_issues.go calls out hold-on-advisory explicitly.
β hive: agent=reviewer backend=copilot model=claude-fable-5.1 copilot=1.0.88
The four Renovate Dependency Dashboards on projectbluefin (common#245, gutenprint-printer-app#45, hplip-printer-app#42, ps-printer-app#50) wear needs-human from a triage bot, so the stage tracker's waiting_on_human block listed them as approve-needed. The monitor never draws a standing meta issue and the stage engine gives it no stage (#16833), but stageTrackerWaitKind did not look at the bucket, so the badge, the topbar pill and the MCP's waiting_on_you counted rows nobody can see or act on. stageTrackerWaitKind now returns no wait for an issue in the dependency_dashboard or hive_advisory bucket, the same structural classification (bot author plus dashboard title, or the advisory title/label) the enumeration and the SPA already use. Claude-Session: https://claude.ai/code/session_01BQYnPGFJx81c8JXR3o1sg1 Signed-off-by: Danathar <Danathar@users.noreply.github.com> Co-authored-by: Danathar <Danathar@users.noreply.github.com> Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Executive summary
Bot Dependency Dashboards were parked at
stage/045-needs-decision. Examples: projectbluefin/gutenprint-printer-app#45, projectbluefin/ps-printer-app#50 and projectbluefin/hplip-printer-app#42, all mergeraptor's. A triage bot addedneeds-human, and the stage engine turned that into a park.A standing meta issue is a control panel or a report, not work. The stage engine now derives no stage for one, in the sweep and out of cycle, so the 045 comes off. #16786 already keeps these issues off the activity monitor's rows. This is the stage label itself.
What
needs-humanstage/045-needs-decisionstage/045-needs-decisionneeds-human, β¦)Why
Enumeration already files these issues in their own work-detail buckets:
dependency_dashboardandhive_advisory(standingMetaIssueKindFor).deriveIssuecarried those buckets through as unknown, but checkedPark.NeedsHumanfirst, so a park label won.How
stageengine:BucketDependencyDashboardandBucketHiveAdvisory.issueStagereturnsNonefor them before any park check, with the reasonstanding meta issue, not work: <bucket>.github/stage_facts_sink.go: the enrichment classifies the fresh read withstandingMetaIssueKindForand sets the bucket. An out-of-cycle fact then derives the same thing as the sweep.Tests
TestDeriveIssueTable(two new rows)TestStandingMetaIssueDerivesNoStage(sweep row, sink fact, enrichment)needs-humankeptgo test ./pkg/stageengine/... ./pkg/stagelabels/... ./pkg/github/... ./pkg/writeback/check-docs-citations.py src/docsNotes
stageFactsServergained optionaltitle/loginfields, so the fixture can serve a bot-authored dashboard.π€ Generated with Claude Code
https://claude.ai/code/session_01BQYnPGFJx81c8JXR3o1sg1