Repository navigation
finding(tooling): the i18n gate's class-8 header still says timeline.relative.* is five abstaining rows — measured 0 since objectui#7887 retired them #8102
Description
Activity
Triage: lands in
domain:devx(scripts/check-i18n-call-site-keys.mjs+ its pin file); rationale: class (a) — a stated count measured false by the gate itself:factoryRowsNoEnKeyis 0, not 5.Bug. Same family as objectstack#16007 / #16200 and objectui#8122, all graded today.⭐ What lifts this above a stale numeral, and it should open the PR: the sentence is not decoration — it is the worked example the header uses to justify printing the abstention count rather than absorbing it, which is the property objectui#7874 exists for. ⇒ A reader checking whether the instrument still has that blind spot is told "five rows, here they are", goes to look, and finds nothing. The two natural readings — "the walk stopped seeing them" or "the prose is wrong" — are indistinguishable from the text, and only one is right. A false example undermines the design decision it was written to support.
priority:p3: prose in a gate header and its pin, ⛔ no behaviour, and the gate's counters are correct.⭐ Route: option 2 — make the prose derivable. Drop the specific count from the header and let the pin file assert the current counters (it already asserts several), so a change in the blind-spot size fails a test instead of quietly falsifying a comment. The card's own objection to option 1 is decisive and is the same lesson this seat has now recorded four times today (objectstack#16007, #16200, objectui#8122, and here): "Fixing the number restores the truth for exactly as long as it takes the next retirement to land." ⇒ The cost of option 2 — a pin that has to be updated deliberately — is the point, not a drawback.
⚠️ Both sites move together or neither does:check-i18n-call-site-keys.mjs's class-8 abstention list andscripts/__tests__/check-i18n-call-site-keys.test.ts's "leaves a key en does not define to class 1" case. ⛔ Fixing one leaves the other asserting five rows that do not exist.⭐ Note the tell already in the tree:
packages/test-support/src/defaults-table-scan.ts's own header already records the objectui#7887 retirement. ⇒ One carrier followed the change and two did not — the "one fact, N carriers, N−1 corrected" shape this seat formalised on objectstack#16198 today, and further evidence that the derivable form is the right answer here.⛔ Correctly not ridden into objectui#7877 / PR #8101 — a stale census number in prose is a different defect from that card's value-compare widening.
⛔ This seat grades and routes only: not claimed, not dispatched, no code.
Generated by Claude Code
Claim: PM loop round R46 —
domain:devx @ objectuiexecution seat.
Session:session_01FhBNJcLRZLe8M87VcUgpKr
Branch:claude/issue-8102-i18n-class8-header
Worktree:/home/user/objectui-issue-8102
Domain:domain:devx
File surface: the class-8 header prose in the i18n gate the card names. ⛔ No key, no locale pack, no assertion. Labelspm:queue→pm:dispatchedand assignee set in one write, read back at 2026-09-07T21:55Z.⚠️ This is the third stale-header card this seat has dispatched today, and the previous two both had a falsified premise inside them — #8006's line numbers had drifted, and #8029's card and its triage comment cited ascripts/placeholder-spelling.mjsthat does not exist, which would have put a new false sentence into a card about a false sentence. ⇒ The dispatch requires the count and the named keys to be re-derived from the gate's own output on today's tree before a single word is rewritten, and requires the dev to say so if the card is wrong about its own subject.Batch independence: disjoint from #8015 (
content/docs/guide/ci-cd-pipeline.md+scripts/__tests__/ci-cd-pipeline-doc.test.ts) and #8113 (the VS Code extension page).
Generated by Claude Code
Dispatched —
domain:devx @ objectuiseat, PM sessionsession_01FhBNJcLRZLe8M87VcUgpKr, R46, 2026-09-07T21:55Z. Three of three slots in flight (with #8015 and #8113).Dev.
os-devsubagent onclaude/issue-8102-i18n-class8-header, worktree/home/user/objectui-issue-8102. Race re-read at 21:56Z: oneClaim:,5575965632.⛔ The instruction that carries this dispatch is re-derive before rewriting, and it is not boilerplate. This is the third stale-header card this seat has sent out today and both predecessors carried a falsified premise inside the card: #8006's line numbers had drifted twice over, and #8029's card and its triage comment cited a
scripts/placeholder-spelling.mjsthat does not exist — copying it would have put a new false sentence into a card about a false sentence. ⇒ Numbers, key names and classification come from the gate's own output on today's tree. If the card is wrong about its own subject, the dev says so; a corrected card with zero implementation is a good outcome.⭐ Also carried: prefer a form that cannot go stale. This shift landed the same lesson twice — a sentence naming a set survives, a sentence restating a count does not, and a claim that declares its own measurement point ("measured at commit X") is a third, acceptable kind.
⛔ No key added, removed or re-spelled; ⛔ no locale pack; ⛔ no assertion edited.
Generated by Claude Code
os-dev-report
{ "issue": 8102, "status": "done", "branch": "claude/issue-8102-i18n-class8-header", "worktree": "/home/user/objectui-issue-8102", "pr": "https://github.com/objectstack-ai/objectui/pull/8421", "pr_number": 8421, "pr_state": "draft", "base": "4b4d35a7d5cf41a537a3841a5cef5abeac5d3559 (origin/main at fetch time)", "head_sha": "fd5aca520588cbdee6609bc1d62006c04e5d9284", "premise_still_valid": true, "card_accurate_about_its_own_subject": { "verdict": "yes on both sites it names, but incomplete, and its own pasted measurement is already stale", "site_1_verified_verbatim": "scripts/check-i18n-call-site-keys.mjs:412 did read 'The 5 such rows on this tree (`timeline.relative.*`, `packages/plugin-timeline/src/useTimelineTranslation.ts`)'", "site_2_verified_verbatim": "scripts/__tests__/check-i18n-call-site-keys.test.ts:1410 did read '`timeline.relative.*` is five such rows on `main` today.'", "claim_that_factoryRowsNoEnKey_is_0": "confirmed by running the gate on the branch point, not inherited from the card", "what_the_card_missed": "the same falsified fact is carried five more times (gate 388, 450, 2016, 2840; pin 1673), all spelling it as 841 comparable rows instead of 846. Correcting only the two the card names would have reproduced the exact 'one fact, N carriers, N-1 corrected' shape the card was filed about.", "cards_own_number_is_stale_too": "the card quotes 841 compared / 841 matching measured at 924edd915; the branch point compares 846 and matches 846. Five unrelated rows were added since. The card demonstrates its own thesis in its own body. This does not falsify its premise about its subject.", "no_ghost_paths": "unlike objectui#8029, every path the card names exists: both files, both quoted sentences, and packages/test-support/src/defaults-table-scan.ts's header recording the objectui#7887 retirement." }, "gate_output_vs_header_claim": { "measurement_command": "node scripts/check-i18n-call-site-keys.mjs on 4b4d35a7d, plus analyze(repoRoot) read directly for the counters the summary line does not print", "measured_today": { "factorySites": 32, "factoryTables": 32, "factoryUnreadableTables": 0, "factoryRows": 846, "factoryComparedRows": 846, "factoryMatchingRows": 846, "factoryRowsNoEnKey": 0, "factoryUnjudgedRows": 0, "factoryUnreadableRows": 0, "handRolledComparedRows": 215, "handRolledTables": 2 }, "summary_line_verbatim": "Factory defaults tables: 32 createSafeTranslation site(s) over 32 distinct table(s) (0 unreadable) - 846 row(s) compared against their en value, 846 matching, 0 on a key en does not define, 0 with no single comparable en value, 0 unreadable.", "carriers_of_the_falsified_fact": [ { "where": "gate:412", "claimed": "The 5 such rows on this tree (timeline.relative.*)", "measured": "factoryRowsNoEnKey = 0", "verdict": "FALSE" }, { "where": "gate:388", "claimed": "846 rows, 841 of them comparable against en", "measured": "846 rows, 846 comparable", "verdict": "FALSE in the second half (846 total is coincidentally still right)" }, { "where": "gate:450", "claimed": "the factory walk compares 841 rows", "measured": "846", "verdict": "FALSE" }, { "where": "gate:2016", "claimed": "a factory walk of 841 rows would otherwise mask a registry that resolved nothing", "measured": "846", "verdict": "FALSE" }, { "where": "gate:2840", "claimed": "factoryComparedRows is 841", "measured": "846", "verdict": "FALSE" }, { "where": "pin:1410", "claimed": "timeline.relative.* is five such rows on main today", "measured": "0", "verdict": "FALSE" }, { "where": "pin:1673", "claimed": "the factory half's 841 rows would hide a registry that resolved nothing", "measured": "846", "verdict": "FALSE" } ], "carriers_already_in_the_durable_form_and_left_untouched": [ { "where": "gate:2827", "text": "factoryComparedRows WAS 841 over 32 tables WHEN THE CLASS LANDED", "verdict": "TRUE - declares its measurement point. This is the local idiom the PR copies." }, { "where": "pin:1487", "text": "The measurement this class SHIPPED ON: 32 sites, 32 tables, 846 rows, 841 comparable, 0 drifted", "verdict": "TRUE - dated the same way" }, { "where": "gate:2019", "text": "objectui#7874 exists purely because the factory half PRINTED its 5 abstaining rows instead of absorbing them", "verdict": "TRUE - past tense about a historical event" } ], "numbers_true_today_that_this_PR_deliberately_did_not_touch": "gate:453 'a measured 215 (80 + 135)' and gate:2845's how-150-was-chosen paragraph both restate a count and both are TRUE today (handRolledComparedRows = 215). They are the same rot MODE but not the same falsified FACT, so they stayed out of scope. Reported rather than edited." }, "what_pins_this_header": { "the_prose": "NOTHING. No test reads scripts/check-i18n-call-site-keys.mjs as text; no check:* script gates gate-header claims (package.json's whole check:* list was read). That is why the sentence rotted in silence.", "the_fact_it_asserted": "CONTRADICTED BY A PASSING TEST. packages/plugin-timeline/src/__tests__/timeline-relative-defaults-retired-7874.test.ts asserts the five timeline.relative.* rows absent, and packages/test-support/src/defaults-table-scan.ts's header records the #7887 retirement. The tree held a green pin and a header asserting the opposite.", "the_counter_behind_it": "UNPINNED, and this is the mechanism of the whole defect. The hand-rolled half's case pins handRolledRowsNoEnKey / handRolledUnjudgedRows / handRolledUnreadableRows / handRolledMatchingRows === handRolledComparedRows. The factory half's case pins only factoryUnreadableTables and factoryUnreadableRows. factoryRowsNoEnKey going 5 to 0 therefore failed nothing. Filed as objectui#8423." }, "files_changed": [ "scripts/check-i18n-call-site-keys.mjs (comment prose only; 6 edits)", "scripts/__tests__/check-i18n-call-site-keys.test.ts (comment prose only; 2 edits)" ], "diffstat": "2 files changed, 56 insertions(+), 33 deletions(-) - comment lines only, zero executable bytes", "staleness_mode_named": { "abstention_bullets": "each names the counter it increments (factoryUnreadableTables / factoryUnreadableRows / factoryRowsNoEnKey / factoryUnjudgedRows) instead of restating its value; all four print on every run. Staleness mode: a RENAMED COUNTER, which is a rename inside the same file and breaks at the identifier, not silently in prose.", "worked_example": "moved to the tense where it is permanently true - the five rows were what only this counter could see, and once it did they were retired in #7887. Staleness mode: NONE; it is a statement about #7874/#7887, which do not move. It is also a stronger argument for printing the count than an occupied bucket was: the empty bucket is the outcome the printing bought.", "shipping_census": "keeps its numbers and declares its measurement point ('the census objectui#7567 SHIPPED ON ... a reading of the tree that card landed against'), then points at the summary line for today's. Staleness mode: NONE; a dated claim about a fixed commit.", "the_three_bare_841s": "become 'hundreds of rows', which is what those sentences actually argue. Staleness mode: fails only below ~100 rows, far under the collapse floor of 500 that exits non-zero eight lines away.", "the_three_sibling_bullets_that_said_0_on_this_tree": "given the same treatment although they were TRUE today. Leaving three bullets restating a count beside one that became derivable would have kept three-quarters of the rot mode inside the very list the card was filed about." }, "tests": "ALL through /home/user/objectstack/scripts/pm/os-verify-lock.sh, parts joined with '&&' so each VERDICT covers every part. (1) 'node scripts/check-i18n-call-site-keys.mjs && node scripts/check-control-bytes.mjs && node scripts/check-changeset-presence.mjs' -> 'VERDICT command-exit 0 - held the lock 8s - waited 186s'. Real output: 'Factory defaults tables: 32 createSafeTranslation site(s) over 32 distinct table(s) (0 unreadable) - 846 row(s) compared against their en value, 846 matching, 0 on a key en does not define, 0 with no single comparable en value, 0 unreadable.' / 'check-control-bytes: OK (scanned 6662 tracked text file(s); skipped 85 binary).' / 'Compared the working tree with 4b4d35a7d (merge-base with origin/main): 2 file(s) changed, 0 of them published source of a package the release covers, 0 of them a manifest whose published contract moved, 0 under a package changesets ignores, 0 changeset(s) added.' + 'No source or published contract of a released package changed in this range, so no changeset is owed.' (2) 'pnpm exec vitest run scripts/__tests__/check-i18n-call-site-keys.test.ts' -> 'VERDICT command-exit 0'; 'RUN v4.1.10 /home/user/objectui-issue-8102' (repo root, root config), 'Test Files 1 passed (1)', 'Tests 137 passed (137)'. (3) 'node scripts/check-comment-mask-corpus.mjs && pnpm exec eslint THE_TWO_FILES' -> 'VERDICT command-exit 0'; corpus sweep '4458 files, 1 disagree ... within the residue objectui#7882 is holding open (ceiling: 1 file(s), 1517 over-masked bytes, 0 fabricated)' - the standing residue, untouched by this PR; eslint silent. (4) 'TURBO_SCM_BASE=4b4d35a7d pnpm exec turbo ls --affected' -> '0 no packages', so no package suite is owed. (5) 'node scripts/check-governed-queue-guard.mjs --test THE_TWO_FILES' -> 'NOT GOVERNED - 2 path(s) checked against 5 governed surface(s); none matched.' (6) grep -naP over both files for control bytes: no hits. ABLATION: none run and none owed - the change is comment-only with zero executable bytes, so there is no behaviour a mutation could show absent; the falsifiable claim here is the MEASUREMENT, and it was taken by running the gate rather than by reading prose (see gate_output_vs_header_claim). NARROWING DECLARED: repo-wide 'pnpm lint' and the rest of the check:* farm are CI's; the local eslint run covered exactly the two changed files.", "changeset": "CHECKED, NOT GUESSED. node scripts/check-changeset-presence.mjs says 'No source or published contract of a released package changed in this range, so no changeset is owed.' No .changeset/*.md added. No skip-changeset label applied - this repo declares 'no release' with an empty-frontmatter changeset when one is owed, and here the gate says none is; the label is not this repo's route.", "labels_written": "NONE. The card's pm:dispatched and assignee were set by the PM and are untouched. The new finding objectui#8423 was created with finding + tooling + domain:devx and no assignee, via the additive REST create-issue endpoint.", "mcp_calls": "0 - every GitHub read and write went through repo-scoped REST (probe: GET /repos/objectstack-ai/objectui/issues/8102 -> HTTP 200). Dedup went through the REST list endpoint (state=all, labels=tooling,domain:devx, 102 issues, control word #8102 present in the listing) plus local grep for factoryRowsNoEnKey|abstention|blind.spot|abstain - no prior card. Card body, comments, open-PR list, labels, PR create, issue create, PR read-back: all REST.", "conflicts_to_flag": [ { "what": "The dispatch prompt and the PM claim comment both say '⛔ no assertion is edited' / '⛔ No key, no locale pack, no assertion'. The triage comment on the card routes to '⭐ option 2 - make the prose derivable. Drop the specific count from the header AND LET THE PIN FILE ASSERT THE CURRENT COUNTERS (it already asserts several), so a change in the blind-spot size fails a test instead of quietly falsifying a comment.' Those two cannot both be obeyed.", "how_it_was_resolved": "The dispatch is later, more specific to this job, and its own star instruction describes prose forms only ('a sentence naming a set survives; a sentence restating a count does not' / 'if the header must carry numbers, say what re-derives them and when they were taken'). I read it as the seat narrowing option 2 to its prose half. So: the prose half is implemented in full, and the pin half is filed as objectui#8423 rather than dropped. I did not pick a side silently.", "what_PM_should_decide": "whether to dispatch objectui#8423 (four lines mirroring the hand-rolled case), or to tell this seat the assertion ban did not cover adding the missing pins, in which case it is a one-commit follow-up on this same branch." }, { "what": "In-place-fix exemption exercised, and the claim's declared file surface is thereby wider than the card's two lines.", "how_it_was_resolved": "All four conditions hold: same defect class (a present-tense census number falsified by the same #7887 retirement), mechanical with the corrected form already demonstrated in the same file (gate:2827), no other claim on either file (checked against all 12 open PRs - none touches scripts/check-i18n-call-site-keys*), and the same gate family with no new verification surface. I could not post a surface increment as a second Claim: comment (the dispatch forbids one), so the increment is declared in the PR body's 'What is in scope' section and here. PM: if the wider surface is unwanted, the extra carriers are separable - they are five one-line hunks.", "what_PM_should_decide": "nothing blocking; this is a declaration, not a question." } ], "four_axis_analysis": { "候选方案": { "A": "修正两处数字(卡片的 option 1):把「5 rows」改成「0」,把「841 comparable」改成「846」。", "B": "分诊路线 option 2 的完整版:散文改为可推导 + 在 pin file 补上 factory 半边的四条断言。", "C (已实施)": "散文侧全量可推导化(抽象项命名其计数器、worked example 改为永真时态、shipping census 日期化、三处裸 841 改为 hundreds),pin 缺口另立 objectui#8423。" }, "实际业务需求": "实测:这条 header 服务的是**读这把尺子的人**(维护者与后续 agent),不是终端用户。谁在写/谁在读,已实测:(1) 该 header 唯一的读者路径是有人打开 gate 源码判断「这个仪器还有没有那个盲区」;(2) 那个判断今天会得到假答案 —— 树里同时存在一条断言五行已退休的 green pin(timeline-relative-defaults-retired-7874.test.ts)与一句断言五行还在的 header,两者互相矛盾;(3) 这条散文不是装饰,它是 header 用来论证「为什么打印抽象计数而不是吸收掉」的 worked example,也就是 objectui#7874 这张卡存在的理由本身。⇒ A 让答案在下一次退休之前为真(卡片自己的原话:「Fixing the number restores the truth for exactly as long as it takes the next retirement to land」),而且**本卡已经亲自证伪了 A**:卡片正文里粘贴的 841 在分支点上已经是 846。B/C 让答案不再依赖任何人记得改注释。这条轴排除 A,不是陪衬。", "项目长远合理性": "北极星是 contract-first、no workarounds。A 是典型的临时补丁:它把同一个欠债推到下一次退休,长期代价是这张卡以同样形态再开一次(本席位今天已记录四次:objectstack#16007、#16200、objectui#8122、本卡)。B 与 C 都符合长远方向,区别在**谁来兑现**:C 让散文不再声明一个它无法维持的事实(声明面收缩到永真的部分),B 在 C 之上再让机器守住数字。严格说 B 才是完整解 —— C 单独存在时,factoryRowsNoEnKey 从 0 变回非 0 仍然只在 CI 日志里出现一行,没有任何东西会红。⇒ 长远轴的诚实结论:C 是正确方向上的一半,#8423 是另一半,不应被读成「已完结」。", "防 AI 写代码犯错(尤其防 AI 写元数据 app 犯错)": "本卡不直接触及元数据编写面,但同一原则可直译:**契约收紧优于消费端宽容**。这里的「消费端宽容」类比就是 A —— 把一个数字抄进注释,让读者(人或 AI)去信任一句无人兑现的声明;AI 读 header 会把它当规格用,而这正是「声明一个运行时不兑现的能力」的注释版本。C 做的是**收紧声明面**:散文不再声明它兑现不了的当期计数,改为指向每次运行都重算的那一行;抽象项命名标识符而不是抄值,于是重命名会在同文件里炸开,而不是静默漂移。B 更进一步,把声明变成机械强制(断言即 publish 时的响亮拒绝)。⇒ 这条轴上 B 最强、C 次之、A 最弱且有害:A 会把一个新的假句子写进一张关于假句子的卡,正是 objectui#8029 的形态。", "创业阶段不扩散需求": "维护者 2026-08-04「先专注核心能力」与 2026-08-27「不考虑渐进、立即退休」两条都适用。C 的成本是零新增机制、零新增验证面、零 changeset、纯注释;它不扩张任何能力面,只是把已有声明缩到能兑现的范围 —— 完全符合从紧默认。B 增加四行断言,也谈不上扩散(镜像的是同文件里已有的 hand-rolled 写法,不是新机制),但它**确实是本卡派发词明令排除的动作面**,在创业阶段「不因为顺手就扩大本轮范围」同样是从紧。至于「过渡不渐进」:C 没有留任何双拼写或分阶段窗口 —— 七处载体一次改完,没有保留一个「暂时还写着 841」的过渡副本,这正是立即退休的写法。⇒ 这条轴支持「本轮交 C、把 B 的增量独立立卡」,而不是支持「只做 A」。", "推荐": "推荐 C + objectui#8423(即 B 拆成两轮)。逐轴理由:业务需求轴排除 A(答案必须对今天的树为真,且不能在下次退休时再假);长远合理性轴要求最终到达 B,但允许拆分,因为 C 单独已经消除了「假句子」这一具体缺陷,而 #8423 处理的是「无人守数字」这一机制缺陷 —— 两者是不同的缺陷,分开落地符合本卡自己「不搭顺风车」的判据;防错轴上 B 最强,所以 #8423 不是可选项而是欠账,报告里明确标注不得读成已完结;创业从紧轴支持本轮不越过派发词划定的动作面。四轴唯一的张力在长远/防错轴 vs 创业从紧轴 —— 前两条想要立刻补上断言,后一条与派发词的 ⛔ 要求本轮止步于散文。张力如实呈上:若维护者或 PM 认为 ⛔ 不覆盖『新增断言』,#8423 可在本分支一个 commit 内完成,不需要新一轮派发。" }, "open_questions": [ { "question": "派发词的 '⛔ no assertion is edited' 是否也禁止**新增**断言?分诊评论的 ⭐ 路线(option 2)明确要求 pin file 断言当期计数,两者直接冲突。", "options": [ "A - 按派发词从严理解(本轮所采):本轮只改散文,pin 缺口作为 objectui#8423 独立派发。", "B - 认定 ⛔ 只禁『修改既有断言』、不禁新增:则在本分支追加一个 commit,镜像 hand-rolled 那四行(factoryRowsNoEnKey / factoryUnjudgedRows / factoryUnreadableRows / factoryMatchingRows === factoryComparedRows),#8423 随即关闭。" ], "recommendation": "A,因为派发词晚于分诊评论、且专门针对本次执行面写就,而其 ⭐ 段落通篇只描述散文形态;同时 B 的价值没有被丢弃,已立卡。若 PM 判定为 B,追加成本约一个 commit,不需要新 worktree。" } ], "out_of_scope_findings": [ "filed as objectui#8423: the i18n gate's factory half pins two of its four abstention counters (factoryRowsNoEnKey and factoryUnjudgedRows unpinned, factoryMatchingRows === factoryComparedRows unpinned), while the hand-rolled half pins all four - which is exactly why factoryRowsNoEnKey going 5 to 0 failed nothing. Labels finding + tooling + domain:devx, no assignee.", "NOT filed, reported only: gate:453 'a measured 215 (80 + 135)' and gate:2845's how-150-was-chosen paragraph restate counts in the present tense and are TRUE today (handRolledComparedRows = 215). Same rot mode, not the same falsified fact, so they were left alone rather than turned into scope creep. If PM wants the whole file converted to the dated/derivable idiom in one pass, that is a deliberate decision to make, not something to smuggle into this card." ], "summary": "Re-derived the gate's numbers by running it on the branch point (4b4d35a7d) instead of trusting the card: factoryRowsNoEnKey is 0, and factoryComparedRows is 846 not 841. The card is accurate about both sites it names but incomplete - the same falsified fact is carried five more times in the same two files, and the card's own pasted measurement (841 at 924edd915) is itself already stale, which demonstrates its own thesis. PR objectui#8421 (draft) converts every carrier of that fact to a form whose staleness mode is named: the abstention bullets name the counter they increment rather than restating its value, the worked example moves to the tense where it is permanently true (the five rows were what only this counter could see, and once it did they were retired in #7887 - a stronger argument for printing the count than an occupied bucket ever was), and the shipping census keeps its numbers while declaring its measurement point, copying the durable idiom already present at gate:2827. Comment-only, zero executable bytes. Nothing pins this header - no test reads the gate as text - while a passing pin in the same repo asserted the opposite fact; the counter behind it is unpinned on the factory half and that gap is filed as objectui#8423." }
Generated by Claude Code
Found while widening that gate's value-compare on objectui#7877 (PR objectui#8101). Not touched there — a stale census number in prose is a different defect from the one that card is about, so it is filed rather than ridden in.
What is stale
Two places state a measured count that is no longer true:
scripts/check-i18n-call-site-keys.mjs, class 8's abstention list: "The 5 such rows on this tree (timeline.relative.*,packages/plugin-timeline/src/useTimelineTranslation.ts) are exactly that — noenleaf and no caller — which is why the count is printed rather than absorbed."scripts/__tests__/check-i18n-call-site-keys.test.ts, in the case named "leaves a key en does not define to class 1, and class 1 cannot see it either": "timeline.relative.*is five such rows onmaintoday."Measured on
924edd915by the gate itself:factoryRowsNoEnKeyis 0, not 5. Those rows were retired in objectui#7887, whichpackages/test-support/src/defaults-table-scan.ts's own header already records — the prose in the gate and its pin file did not follow.Why it is worth a card rather than a silent edit
The sentence is not decoration. It is the worked example the header uses to justify printing the abstention count instead of absorbing it — the property objectui#7874 exists because of. A reader checking whether the instrument still has that blind spot is told "five rows, here they are", goes to look, and finds nothing; the natural readings are "the walk stopped seeing them" or "the prose is wrong", and only one is right.
It is also an instance of a class this lane keeps meeting: a census number transcribed into prose, with nothing that fails when the tree moves under it. Fixing the number restores the truth for exactly as long as it takes the next retirement to land.
Two shapes, for triage rather than as a recommendation
No assignee, per the lane's rules.
Filed by the
domain:devx @ objectuiseat while implementing objectui#7877, sessionsession_01FhBNJcLRZLe8M87VcUgpKr.