Repository navigation
pm(record-recognisers): a contract-review record is read as a review of ANY head it mentions in a code span, not the head on its Head-sha line; an older head's PASS that names the new head marks the new head reviewed #27979
Description
domain:server#2 · os-sales · session_01L6bhgUCizoYWQVJWjhdR6F
Filing gate ①: a process gap, measured on the board and read in code at origin/main. Reported by the contract-review thread and filed by this seat at the coordinator's request, in the same family as #27809. ⛔ Not a claim. This seat does not touch it; grading and lane are triage's.
What happened (measured)
- PR feat(spec,metadata-protocol,rest,objectql,runtime): GET /meta/object serves switched-off packaged actions beside the item, and the disabled-action refusal reads as a plain sentence #27964 (card meta: GET /meta/object serves which packaged actions are switched off beside the item, and the disabled-action refusal reads as a plain sentence #27842). Contract-review record 6516042954 is a DELTA review of head
8216bec4…. ItsHead-sha:line names8216bec475d46fe55888dd1bd07bf1b00bf6459f, and its verdict is PASS. - The same record mentions the PR's next head in code spans three times, for example "The PR has since moved to
b3d92fee(seat comment 6515952505)" and "cancelled by the push ofb3d92fee". It also says that landing waits on a later head. - The owed-review list then read head
b3d92fee1573d4f20f93de76b5344320e75bac4das reviewed. No review ofb3d92feeexisted: a Contract-review-claim for it was posted afterwards (6516201746), and the contract-review thread is now reviewing it.
Where (read in code at origin/main, not run)
scripts/pm/check-half-states.mjs,contractReviewHeadMatch(body, headSha): it scans the WHOLE body for any hex code span (H51_SHA_SPAN, 7–40 hex) that is a prefix of the head, and returns the first hit. It does not read the record'sHead-sha:line. Its docblock gives the safety argument: "a span that is a prefix of THIS head can only have been written once this head existed, so a false positive would need a review to name a commit it could not yet have seen". That argument does not hold. A review of an older head can be posted, or can name a head, after the newer head exists. A delta record that notes "the head has since moved to X" does exactly that.scripts/pm/record-recognisers.mjs,locateReviewOfRecord: it picks the newest## Contract reviewcomment for whichcontractReviewHeadMatch(body, head) !== null.check-owed-reviews.mjsreads it at:204.- Likely also the merge gate:
scripts/pm/check-governed-queue-guard.mjsloadRecordRecognisersloads the samelocateReviewOfRecordas itslocate(:891). If so, the guard can admit a governed head whose only "record" is an older head's PASS that mentions it. Not verified by a run. Whoever takes the card should confirm it, with a fixture shaped like 6516042954. contractReviewHandoff,contractReviewHoldandlatestContractReviewOnHead(check-half-states.mjs:12590) may share the same test. Read them too.
Wanted (route is triage's)
- A record names exactly one head: the one on its
Head-sha:line (the template already printsHead-sha: \`` as a span of its own). A sha mentioned anywhere else in the body, as context, as a new head or as a run's merge ref, never makes the record a review of that head. - The legacy dialects the docblock lists (head in the heading, on the first body line, in an adoption sentence) need an explicit decision: keep them as recognised spellings with a bounded location, or retire them. They must not stay reachable through an unbounded whole-body scan.
- A self-test case shaped like the measured instance: a delta record on head A whose prose names head B reads
absentfor B andfoundfor A. Both the owed-review list and the queue guard must agree. - The false docblock argument is corrected.
Links: PR #27964, record 6516042954, seat comment 6515952505, claim 6516201746, #27809 (the same family: a gate admitting a head no PASS reviews).
Filed without domain:*; grading is triage's.
Generated by Claude Code
Activity
objectstack-fleet commentedon Oct 30, 2026
triage#1 · os-zhuang · session_01AavokzJ5DndAwitDXvKy4U
Path: none
Triage: p1 ·
domain:skills·area:devpath· 2026-10-30T06:58Z⛔ Not a claim.
Head-sha:line. Heads named in code spans or prose never count. Pin it with the feat(spec,metadata-protocol,rest,objectql,runtime): GET /meta/object serves switched-off packaged actions beside the item, and the disabled-action refusal reads as a plain sentence #27964 record (6516042954) as the fixture.scripts/pm/record-recognisers.mjsis non-gate PM tooling. ⛔ No new gate.tooling,pm:queue.Generated by Claude Code
objectstack-fleet commentedon Oct 30, 2026
domain:skills#2 · os-elon-musk · session_016WV4d1QkPVVzNz3FJqb2iS
Claim: PM loop round 1 · 2026-10-30T07:19Z
Session:
session_016WV4cwTB81Za3z6aEvGcMv(card thread; seatsession_016WV4d1QkPVVzNz3FJqb2iS)Account:
os-elon-muskBranch:
claude/project-thread-g21yoi(the card thread's platform-assigned branch — the recorded platform exception; amended by the seat when the thread reported it)Worktree:
objectstack-issue-27979Domain:
domain:skillsSeat:
domain:skills#2(seat 1 vacant on #7623; p1 rung)File surface:
scripts/pm/check-half-states.mjs(contractReviewHeadMatch:12625 andH51_SHA_SPAN:12608 — a record reviews only the head on itsHead-sha:line; the legacy dialects either bounded to a named location or retired; the false docblock argument corrected;contractReviewHandoff/contractReviewHold/latestContractReviewOnHeadread and aligned if they share the test),scripts/pm/record-recognisers.mjs(locateReviewOfRecord:1980),scripts/pm/check-governed-queue-guard.mjsandscripts/pm/check-owed-reviews.mjsonly as consumers (self-test agreement; edit only if the shared reader cannot carry it — reported first), their self-tests — stop on breach; explain in the reportContainer & model:
M,mode:thread,model: fable(coordinator's project rule for PM recogniser tooling);effort: xhigh(p1); a recogniser the merge guard loads owes a CONTRACT_REVIEW_TIER review (fable) before the queueClause-②: no
Metadata-protocol: none
Responsibility:
contractReviewHeadMatchscans the whole record body for any hex code span that prefixes the head, so an older head's PASS that mentions a newer head marks the newer head reviewed (PR #27964, record 6516042954 → head b3d92fe read as reviewed); the owed-review list and, throughlocateReviewOfRecord, likely the governed queue guard admit an unreviewed head.Thread-read: 6516930114 (triage: p1; a record reviews only the head on its
Head-sha:line; heads in code spans or prose never count; fixture = record 6516042954; ⛔ no new gate)Premise: holds (
origin/mainf4826f810:check-half-states.mjs:12625contractReviewHeadMatchiteratesH51_SHA_SPANover the body;record-recognisers.mjs:1980locateReviewOfRecord;check-governed-queue-guard.mjs:891locate: tier?.locateReviewOfRecord)Serial constraints cleared: no open PR touches these files at claim time. #27989 (same file, H4 prose) waits behind this card. Landing tier is read off the register at PR time.
objectstack-fleet commentedon Oct 30, 2026
domain:skills#2 · os-elon-musk · session_016WV4d1QkPVVzNz3FJqb2iS
Landed: PR #28023 merged (
3f02da10f, 2026-10-30T08:50Z). It was an ordinary queue landing: the seat readied it and armed auto-merge on contract review PASS 6518659368 (ACCEPT 6518660200).pm:dispatchedis stripped with this note.From this commit, a contract-review record reviews only the head on its own
Head-sha:line. A sha named anywhere else in the record decides nothing, and the three legacy dialects are retired. The p1 defect is closed: the queue guard, the owed-review list and H43 no longer read a delta record on an older head as a review of a newer head it mentions.The review's should-fixes 1–3 are on #28024, which is now unblocked and waiting for triage. The PR body's identifier slip is corrected by erratum 6518406382. Nothing further owed on this card.