Repository navigation
🐛 Activity Monitor: re-review waits on the hive reviewer; CI caption from a full check list - #16636
Conversation
…ull check list A pushed head whose checks are green and which carries no findings or unrebased conflict owes nothing to its author: the re-review is the hive reviewer's. The row says so (rework.owed_by = hive-reviewer / author), the pill reads 'waiting on 🐝 hive reviewer · re-review of <sha7>', and the Review chevron is the lit one with CI an icon-only green ring. The stale 'CI 7/11 · test not reported' caption came from the GraphQL PR batch: statusCheckRollup.contexts(first: 50) had no totalCount and a truncated page was cached as the head's complete listing, so the required aggregates past the page read as never reported. The query now asks for 100 with totalCount, and a truncated rollup is treated as no listing so enrichPRCI falls back to the paginated REST check-runs. Fixes #16635 Signed-off-by: Andrew Anderson <andy@clubanderson.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 pr_batch_graphql.go, stage_tracker_rework.go, stage_tracker_build.go and the two JS files at 15e41e7 against v5, plus the cache consumer. This looks correct to me.
- Truncation guard.
graphQLContextsTruncated(pr_batch_graphql.go:374-380) comparestotalCounttolen(nodes); on truncationgraphQLCheckRunsreturnsok=false(:321) sostoreGraphQLPRBatchNodenever populatesprBatchCheckRunsfor that key (:277-289) andgraphQLBatchCheckRunsObservationmisses (:392), which is the REST fallback. No stale partial listing can linger:clearGraphQLPRBatchReporuns before each repo batch (:166,:178). Fixtures withouttotalCountdecode to 0 and stayok=true, covered by thelegacycase inTestGraphQLCheckRunsTruncatedContextsReportsNotOK. - Owed-by.
stageTrackerReworkOwedBy(stage_tracker_rework.go:102-116) is applied after CI/review activity are derived (stage_tracker_build.go:559), so the fields it reads are populated. The JS fallback for older servers (tracker-words.js:347-355) implements the same three author-owes conditions in the same order, so a mixed-version hub/spoke renders the same pill.
CI on this head: all required checks green (only prow tide pending). Nothing blocking from my side.
— 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 |
|
Hive status — updated in place; edits do not notify. Full history is in the Activity Monitor. @clubanderson this PR is held at Why:
Next: the hive does not push to your branch. Push a new head (or reply here); CI and the review swarm re-run automatically and the PR returns to the normal path. |
Fixes #16635
What the operator saw (#16615, @thatguywob's clanker)
waiting on 🤖 @thatguywob · re-review pending· CI chevron ACTIVECI 7/11 · test not reported· Review icon-only. Ground truth: head d88928e, every required check green (only prowtidepending), hive's reviewer last reviewed the previous head 6384f9b. Nothing was waiting on the author.Root cause of the stale CI caption
graphQLPRBatchQueryasksstatusCheckRollup { contexts(first: 50) }with nototalCount, andgraphQLCheckRunsreturnedok=trueon a truncated page. d88928e had 71 check runs + 2 statuses (thetestmatrix alone is ~30 jobs), so the required aggregatestest,golangci-lint,gosec,govulncheckfell past the page;enrichPRCItrusted the cache (source: github-graphql-cache+…) instead of its paginated REST fallback and counted them as never reported → 7/11, blockertest, statemissing.Changes
stage_tracker_rework.go,tracker-words.js):rework.owed_byishive-revieweron a pushed head unless the head itself asks the author for something (CI failing on it, changes requested on it, a conflict nobody in the hive is rebasing) →author. The pill readswaiting on 🐝 hive reviewer · re-review of d88928ewith no author presence glyph; the author pill (with glyph) stays only when they owe. Older payloads without the field are read from the activities the same way.pr_batch_graphql.go):contexts(first: 100) { totalCount … }; a truncated rollup is "no listing" soenrichPRCIfalls back to the paginatedListCheckRunsForRef. Prowtideis already a meta check (isMetaCheck) and never gates the verdict.tracker-row-strip.js): a hive-owed re-review is a new-head wait, not a park on a person; with CI green the stalled-track rule lights Review (Review 0/5 · re-review of d88928e) and CI is the icon-only green ring; with CI still running, CI is lit and Review the glyph. Wordingre-review pending (new head)→re-review of <sha7>.Scoped to attribution + staleness; the CI running-vs-passing classification in
stage_tracker_activity.gois untouched.Tests
TestStageTrackerReworkOwedBy16615,TestGraphQLCheckRunsTruncatedContextsReportsNotOK,TestStageTrackerReworkJSupdated.#16615cases instage_tracker_presence.test.js(pill actor = hive reviewer, author keeps the pill only when owing) andstage_tracker_activity.test.js(Review active, CI icon-only green, exactly one active track).node --test src/pkg/dashboard/*.test.js(313 pass),go test ./pkg/dashboard/(full),go test ./pkg/github/ -run 'GraphQL|PRBatch',make dashboard-gen-checkall green.