main-red (intermittent): Test (coverage shard 6/8) times out on body-dialect-census.test.ts:288 ("the platform SHIPS the dialect it is being asked to refuse") at the 60 s test timeout — 4 of the last 12 completed CI runs on main read red on it

Filing gate ①: a product defect with a measured reproduction. It is a red gate on objectui main, intermittent. Reader who acts: objectui domain:devx seat 1 (seat post #5748; scripts/__tests__/ and the CI shard layout are this lane's), in one PR. Filed by that seat · os-warren · session_01FM29ZqLTrnPNXxijioZwMp, from objectstack's PM review of objectstack-ai/objectstack#24829, the console pin move, where it surfaced in the pin's neighbourhood. ⛔ Not a claim. The lane is self-labelled under rule B; grade and area:* are triage's.

What is red

ci.yml's Test (coverage shard 6/8) fails with one annotation, every time:

scripts/__tests__/body-dialect-census.test.ts:288 Error: Test timed out in 60000ms.

The case at :288 is it('the platform SHIPS the dialect it is being asked to refuse', …). The aggregate Test (coverage) then fails with it.

Measured, completed CI runs on main, newest first, read 2026-10-20 ~09:35Z: 12 runs, 4 red.

event	head	conclusion
push	`3cec0901fe`	success
push	`05e3f34b45`	failure (job 118541041662)
push	`e81a37cfd7`	success
push	`c9cb89662a`	success
push	`5115c1f08e`	failure (job 118538066544)
schedule	`768960e386`	failure (jobs 118536154473, 118546258071)
push	`768960e386`	success
push	`7e62822253`, `c209c933d0`, `aaa1296f94`, `9840695f67`	success
push	`f7aac4ce7a`	failure (job 118524668810)

Earlier the same day: 186b78d8f347 (push and schedule) and 91b658a75d64 (push) failed on the same line.
The same commit, 768960e386, passed on push and failed on schedule. So the cause is the case's wall time under coverage instrumentation, not a commit's content.
The test file has not changed since d758f2f2 (2026-10-10, refactor(plugin-list,plugin-dashboard,app-shell): one authority each for ViewSwitcherProps, RecordDetailDrawerProps and PageHeaderComponentProps (objectui#6349, batch 8) #12074). The time grows with what the census reads, not with the test.

Why it matters

Each red push run marks a main commit red. The objectstack console pin rule takes "the newest objectui main commit … green at the move", and PR objectstack-ai/objectstack#24829 had to skip a red tip for exactly this reason. A main red is also this lane's first work (seat post #5748).

Premises, each with a re-check

gh api repos/objectstack-ai/objectui/actions/workflows/ci.yml/runs?branch=main&status=completed&per_page=12 → at least 1 failure, whose shard-6/8 annotation names body-dialect-census.test.ts:288 and the 60000 ms timeout.
git log -1 --format=%h origin/main -- scripts/__tests__/body-dialect-census.test.ts → d758f2f2 (the file is unchanged).

Wanted

The case finishes well inside its budget under coverage on every shard run, with the claim it pins unchanged:

make the census read cheaper (scan once per suite, or narrow the walk), or
give that one case a measured, stated timeout, or
move it to a shard with headroom.

The lane chooses, with its measurement in the PR body. ⛔ Not skip, .todo or a quarantine, and the census claim is not weakened. Measure the case's wall time under --coverage before and after.

Dedupe

MCP search_issues, objectui, open and closed: "body-dialect-census test timed out 60000ms coverage shard main red", 10 hits. None names this case. The nearest are closed: #12256 (a coverage shard at its 40-minute job ceiling), #9271 (a 15 s timeout blinding the coverage gate), #12384 (shards cancelled at the 20-minute job limit) and #12478 (a vi.waitFor flake). All are different cases or limits.

Dedupe words: body-dialect-census timeout coverage shard 6/8; the platform SHIPS the dialect it is being asked to refuse; intermittent main red coverage.

Generated by Claude Code

Claim: PM loop round 24 (objectui seat) · a p2 domain:devx card, an intermittent main red, graded by triage; serial dispatch under the maintainer's 「串行派发」
Session: session_01FM29ZqLTrnPNXxijioZwMp
Account: os-warren (the seat's linked user as GET /user answers it; the card's assignee)
Branch: claude/issue-13210-census-coverage-timeout
Worktree: objectui-issue-13210
Domain: domain:devx
Seat: domain:devx#1 (objectui, seat post #5748)
File surface: scripts/__tests__/body-dialect-census.test.ts, and the census helper(s) it reads, if the fix makes the census cheaper there. The case at :288 ("the platform SHIPS the dialect it is being asked to refuse") must finish well inside its budget under --coverage, with the claim it pins unchanged. Stop on breach and explain in the report.

⛔ No skip, .todo, quarantine, or weakened census claim.
⛔ No .github/workflows/** (Tier H governed).
⛔ No hand edit of scripts/test-shard-timings.json (generated; open bot PR chore(test): refresh the per-file test timings (objectui#12208) #13028).
Container & model: S (one slow test case under coverage, measured before and after), mode:subagent, model: TIER_DEFAULT
Clause-②: no
Responsibility: objectui main's Test (coverage shard 6/8) times out on body-dialect-census.test.ts:288 in 4 of 12 recent CI runs, so main commits read red at random | none: each red push run marks a main commit red, and the console pin rule had to skip a red tip | every objectui main commit, and objectstack's next console pin move
Thread-read: none
Serial constraints cleared: no open objectui PR touches scripts/__tests__/body-dialect-census.test.ts or the census helpers. PR chore(test): refresh the per-file test timings (objectui#12208) #13028 touches only scripts/test-shard-timings.json, which is outside this claim (read at 2026-10-20T11:37Z). No dev of this session is in flight.

Generated by Claude Code

os-dev-report

{
  "issue": 13210,
  "status": "done",
  "branch": "claude/issue-13210-census-coverage-timeout",
  "pr": "https://github.com/objectstack-ai/objectui/pull/13250",
  "session": "session_01FM29ZqLTrnPNXxijioZwMp — mode:subagent, so this is the parent seat's harness-stamped id (the Claude-Session trailer on every commit of the branch)",
  "premise_still_valid": true,
  "summary": "The premise held: the case is a cost close to its budget, not a hang. What pushed it there was found and removed. Since 2026-10-19 (objectui#12957, objectui#13008) both tree walks also read .objectstack-pin/, the git-ignored pinned @objectstack build: 1,879 files, 70 MB, one of them a 24.9 MB JSON Schema file. Under v8 coverage the per-character parse of all that took the case to 37.7 s locally, and the CI coverage shard to its 60 s limit. Four changes: (1) both census helpers add .objectstack-pin to SKIP_DIRS, beside .objectui-tmp; (2) census() runs scanNodes only on a text where mayHoldWantedNode finds a type key with a quoted population key as its value, a necessary condition read off the scanner and tested on the text scanNodes reads; (3) lineCounter builds one newline index per file instead of recounting newlines per hit; (4) two new pins, for the pre-filter's superset property and a two-sided .objectstack-pin skip covering both instruments. The claim, its assertions, its producers and its 60_000 budget are unchanged. Outside the pin, the hit set is byte-identical by digest for three populations, and readers, producers and unchannelled are unchanged. Files changed: scripts/body-dialect-census.mjs, scripts/body-dialect-producer-scan.mjs, scripts/__tests__/body-dialect-census.test.ts. No changeset is owed (check-changeset-presence). Commits 4100bd59c, b2a22411b, 5aef02375 on BASE 7c05f210d. Main has since moved one commit and touched none of these files.",
  "tests": "Machine: cloud container, 4 vCPU Intel Xeon @ 2.10 GHz, 15 GB, Node v22.22.0, vitest 4.1.10. The box is shared: the verify lock excludes other locked runs only. Command: pnpm exec vitest run --coverage --testTimeout=600000 --reporter=verbose scripts/__tests__/body-dialect-census.test.ts, with the four --coverage.thresholds.* set to 0; 'alone' adds -t 'the platform SHIPS the dialect'. WALL TIME OF THE CASE UNDER COVERAGE: inside its file, BASE 7c05f210d 38402 / 37559 / 37692 ms, HEAD 5aef02375 13105 / 13311 / 13915 ms. Alone, BASE 38621 / 39444 / 38480 ms, HEAD 13470 / 14995 / 13110 ms. Whole file (vitest Duration), BASE 39.47 / 38.52 / 38.79 s, HEAD 14.06 / 14.32 / 14.81 s. Uninstrumented, one run: 15694 ms to 7629 ms. That is about 2.8x faster under coverage, and the case now uses 22-25% of its 60 s budget locally. pnpm test:coverage with the shard job's flags (--reporter=default --reporter=github-actions --testTimeout=60000, thresholds 0) on the file: case 13936 ms, 24 passed, exit 0. WHERE THE TIME WENT (BASE; node calling each instrument, NODE_V8_COVERAGE gives the same precise block coverage as vitest's v8 provider): census(root) 9.93 s plain, 25.72 s under coverage, 9.42 s of it under .objectstack-pin (6 hits, 0 body). scan(root) 6.22 s plain, 11.87 s under coverage, 0.90 s of it under the pin (29 hits, all bucket other). The CPU profile under coverage puts scanNodes at 79% of the census and readFileSync plus stat under 4%, so the cost is parsing, not walking. In the scan, walkFrames + scanSource + blank take 69% and lineOf 8%. Hypothesis 1 held in part: parsing costs, walking does not, and beforeAll would remove nothing. Hypothesis 2 was falsified: the other 21 cases take 0-2 ms each (the CLI case 142 ms, on a temporary root). DIGESTS, BASE instruments vs HEAD instruments on one tree, sha256 of hit rows with line numbers: census default population 1586 hits outside the pin, 2c4f7f512b6101b3 on both sides; keys text,image,icon 5586 hits, dfd07ab48a90b991 on both; keys page,card,badge 757 hits, 99cdc37414e33bce on both. scan hits outside the pin equal; readers equal (4 reads, receiver schema); producers 0/0; unchannelled 6/6; literalCarried lit control 106 to 104, still lit. ABLATIONS through ../objectstack/scripts/ablation-replace.mjs in wrap mode. Each mutation landed (anchor 1 to 0, blob e54a96b1ecaf changed). Each restore was proven: blob equals HEAD, git diff HEAD empty, git status --porcelain empty. (a) card.tsx defaultProps gains children with a { type: 'card', body: [...] } node: the case goes red with 'expected [ Array(1) ] to deeply equal []' on the app-metadata assertion, the one row naming the planted card.tsx site. (b) card.tsx defaultProps gains body: [...] with no sibling type, which the census cannot see: the case goes red with 'a producer is BACK ...: expected 1 to be +0'. (c) '.objectstack-pin' removed from the census SKIP_DIRS: the new skip pin goes red, 'expected true to be false'. (d) key escaping removed from mayHoldWantedNode: the new pre-filter pin goes red on { \"type\": \"x+y\" }. No rebuild leg was needed: the tests read the .mjs sources directly, with no dist.",
  "gates": "Local, exit codes captured before any pipe: pnpm test on body-dialect-census.test.ts and body-dialect-producer-scan.test.ts, 2 files, 50 passed, exit 0. pnpm test:coverage with shard flags on the census file, 24 passed, exit 0. pnpm run type-check:scripts, exit 0; --listFiles shows both .mjs files and the test file in the program. pnpm exec eslint --max-warnings=0 on the 3 files, exit 0. check:new-line-citations, 'VERDICT new-cross-file-line-citations: 0 new citation(s)', exit 0. check:control-bytes OK, exit 0. check:test-path-roots OK, exit 0. check:entry-guard, exit 0. node scripts/check-changeset-presence.mjs, 'no changeset is owed', exit 0. check-governed-queue-guard --test on the 3 paths, NOT GOVERNED. The control-byte self-scan of the 3 files found 0. No --ran reconciliation: dispatch-gates is objectstack-only, per the dispatch, so objectui's subset was derived by hand. PR CHECK-RUNS on head 5aef02375, read once at report time: 32 total. Success: Build Docs, Changeset Claim Re-read, Changeset Declaration, Changeset Fixed Group Check, Governed Surface Queue Guard, Line Citation Gate, label. Skipped: Test (coverage shard MATRIX.SHARD/8), Test (coverage), dependabot. In progress: Test (shard 1/16) through (shard 16/16), Lint, Type Check, Repo Gates, Test (dist pins), Build & E2E, Live E2E (informational). NOT MEASURED: CI shard 6/8 job time on the PR. Reason: test-coverage runs only on push and schedule, and its check-run reads skipped on this PR. The first CI reading is the push run on main after merge, which is for the PM to cite.",
  "mcp_calls": "0 — no MCP GitHub calls",
  "api_writes": "2 relay strokes (POST /repos/objectstack-ai/objectstack/dispatches each, executed as objectstack-fleet[bot]): (1) pr_create, i.e. POST /repos/objectstack-ai/objectui/pulls (draft) plus POST /repos/objectstack-ai/objectui/issues/13250/assignees [os-warren], read back byte-identical (9476 bytes); (2) this os-dev-report comment, POST /repos/objectstack-ai/objectui/issues/13210/comments via post-stamped.mjs. Also git push on the branch (5 pushes; not REST). Reads only otherwise (single-card REST reads of the card, the claim, 4 precedents, runs/jobs/check-runs).",
  "deviations": [
    "Ruling 4's second reading, CI's shard 6/8 job time on the PR, does not exist: the coverage lane runs on push and schedule only. Recorded as NOT MEASURED. Job-log downloads for earlier main runs were refused by the container proxy (403), so no CI per-case baseline is quoted either.",
    "Ruling 2: cheaper reads were chosen (narrowed walk, pre-filter, line index). The per-case timeout was not raised, and beforeAll did not apply because no other case walks the tree. Judgment call, declared: the .objectstack-pin skip changes what the instruments count. It drops filesScanned and removes 6 census rows (0 body) and 29 scan rows (bucket other). None of those rows was a body hit or a shipped-source row, and the directory had been inside the walk only since 2026-10-19. So the claim's assertions and producer population are unchanged.",
    "Two new pin tests were added in the test file, which is inside the file surface. They are unit pins of the new code, not a new gate.",
    "Ablations used objectstack's scripts/ablation-replace.mjs, because objectui has no copy.",
    "Commit trailers are the model-free pair per objectui AGENTS.md, not the harness's model-named Co-Authored-By. The PR body ends with os-dev's session-URL footer form, not the harness's suggested lines."
  ],
  "files_changed": [
    "scripts/body-dialect-census.mjs",
    "scripts/body-dialect-producer-scan.mjs",
    "scripts/__tests__/body-dialect-census.test.ts"
  ],
  "open_questions": [],
  "out_of_scope_findings": [
    "carrier: none (承接者:无) · noted, not filed. Since 2026-10-19, any objectui script that walks from the repo root and descends dot-directories may also read .objectstack-pin/. Only this census was measured; other sweeps (for example the repo-sweeping scripts/__tests__ rows the coverage lane's --testTimeout comment names) were not measured. This is an only-read inference, so it is in the PR's Acceptance notes and nowhere else. Dedupe words: objectstack-pin tree walk SKIP_DIRS; repo sweep reads pinned build; coverage lane sweep timeout.",
    "carrier: none (承接者:无) · noted, not filed. The producer scan's CANDIDATE_FILE docblock carries a stale count ('1852 files contain the substring body, 500 satisfy this predicate'); this branch measured 966 candidates. Nothing recomputes the figure (AGENTS.md rule 9). It is a stale prose count, not a defect class, so it is in the PR's Acceptance notes and nowhere else."
  ]
}

Generated by Claude Code

Landed · objectui domain:devx seat 1 · session_01FM29ZqLTrnPNXxijioZwMp · 2026-10-20T13:10Z

reading	result
merge queue	none left: the entry has drained
PR object	PR #13250 merged at 2026-10-20T12:43:01Z as `e537636300` (not governed), on this seat's ACCEPT `6290861067`
objectui `main`, by content	`scripts/body-dialect-census.mjs`, `scripts/body-dialect-producer-scan.mjs` and `scripts/__tests__/body-dialect-census.test.ts` are byte-identical to the PR head `5aef023752`
The red this card names, on `main`	`CI` push run 39424437970 at `e537636300`: success. `Test (coverage shard 6/8)` success (job 118633284867, 2026-10-20T12:43:07Z–2026-10-20T13:01:40Z), all eight coverage shards success, and the aggregate `Test (coverage)` success

State change: PR #13250 closed this card through Fixes #13210 (completed). pm:dispatched comes off in this act.

One green run is the first reading, not the proof, for an intermittent red (4 of 12 runs before the fix). The case measured 13–14 s under coverage after the fix, against 37–39 s before, which is 22–25% of its 60 s budget, so the margin is now structural rather than lucky. If the shard reds again on this line, it is a new finding, not this card.

Noted from the dev's report, not filed:

other repo-root sweeps may also descend into .objectstack-pin/; only this census was measured;
the producer scan's CANDIDATE_FILE docblock carries a stale file count.

Both are in PR #13250's Acceptance notes.

Generated by Claude Code

Navigation Menu

Repository navigation

main-red (intermittent): `Test (coverage shard 6/8)` times out on body-dialect-census.test.ts:288 ("the platform SHIPS the dialect it is being asked to refuse") at the 60 s test timeout — 4 of the last 12 completed `CI` runs on main read red on it #13210

What is red

Why it matters

Premises, each with a re-check

Wanted

Dedupe

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

main-red (intermittent): Test (coverage shard 6/8) times out on body-dialect-census.test.ts:288 ("the platform SHIPS the dialect it is being asked to refuse") at the 60 s test timeout — 4 of the last 12 completed CI runs on main read red on it #13210

Description

What is red

Why it matters

Premises, each with a re-check

Wanted

Dedupe

Activity

objectstack-fleet commented on Oct 20, 2026

objectstack-fleet commented on Oct 20, 2026

objectstack-fleet commented on Oct 20, 2026

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions

main-red (intermittent): `Test (coverage shard 6/8)` times out on body-dialect-census.test.ts:288 ("the platform SHIPS the dialect it is being asked to refuse") at the 60 s test timeout — 4 of the last 12 completed `CI` runs on main read red on it #13210