[finding] check-widening-tells T4 把 ERROR_CODE_LEDGER 的出处行读成扩面 —— 并集实测 282→282、added/removed 皆空,却恒逼出假的 Clause-②: yes

Path: none | instrument (scripts/pm/check-widening-tells.mjs T4 — the why sentence :3903 on ERROR_CODE_LEDGER provenance rows; the accept-set limb only) | family with #19221 (same tool); region-parallel with PR #19438 (T2) | graded by domain:skills#2

由 domain:services 席(session_01AhQASwqJr2Z7XfGWUdvnbF)立卡,证据为本席独立复测,⛔ 非转述。⛔ 未分诊、⛔ 无 domain:*、⛔ 无定级。

缺陷

ERROR_CODE_LEDGER 的分包数组按该文件自己的头部自述是 provenance(出处)不是 identity(身份):「a code emitted by several packages is listed once per emitting package」。因此给已存在的码增加一个发射方包的行,不会让运行时接受的码集变大。

但 scripts/pm/check-widening-tells.mjs 的 T4 把它读成扩面:

✗ T4 packages/spec/src/api/error-code-ledger.zod.ts:1121 —
   a new registration in a registry / catalog — what the runtime accepts grows with no schema file moving

复现(本席跑的,带对照)

树 72d68b905(PR #19437),对照基线 origin/main:

$ git diff origin/main...HEAD > d
$ node scripts/pm/check-widening-tells.mjs --declaration no --diff d
✗ T4 … error-code-ledger.zod.ts:1121 ;  exit 4

而该 diff 对去重并集的实际影响,本席单独测过(⛔ 只计数组元素行,排除注释行 —— 新增注释里恰好写了那个码名,把它计进去就是自己制造答案):

	去重并集
改动前(`origin/main`)	282
改动后	282
added / removed	`[]` / `[]`
对照:`UNIQUE_VIOLATION` 改动前是否已在集内	是

⇒ 接受集一个码都没多。T4 的判据文字(「what the runtime accepts grows」)在这个文件上不成立。

后果

check-clause2-carriers --pair 因此退 4(行 C5:申报 no 而 diff 带 tell)。⇒ 今后每一条 ledger 出处行都会逼出一个假的 Clause-②: yes,或者逼出一次人工豁免。 而 lanes/spec.md 的常设承诺明文预批了跨车道 ledger 追加这一整类 —— 预批的类,却在门禁上恒红。

修法(门禁自己点名的那条)

该工具的失败文本逐字写着:「if the tell is FALSE, repair it here in the matcher (scripts/pm/check-widening-tells.mjs, with a --self-test case pinning the shape), or file that repair as its own card when it is out of this PR's scope」。本卡即它点名的那张卡。 修法要点:T4 对 ERROR_CODE_LEDGER 需按并集判定而非按行判定,并配一个 --self-test 用例钉住「加一条既有码的出处行 ⇒ 不是 tell」。

Seam: consumer: none(席位工具面,无运行时/渲染消费者)

查重词

widening-tells T4 ledger provenance false positive · clause2 C5 error-code ledger dedupe · error code provenance row is not identity · check-widening-tells registry entry dedupe · T4 union unchanged 282

Generated by Claude Code

⚠️ 重要限定:本卡 ⛔ 不意味着「ledger 出处行不触条款②」

_{domain:services seat · session_01AhQASwqJr2Z7XfGWUdvnbF · 写于 2026-09-20T22:14Z。本条由交付 #19307 的 dev 指出缺口后补上,本席核实后同意。}

本卡的主张有一个精确边界,立卡时我没写清,容易被反向误读:

✅ 本卡主张的是:T4 的那句推论(「what the runtime accepts grows」)在 ledger 出处行上为假 —— 接受集实测 282 → 282,added/removed 皆空。这一条成立且不撤。
⛔ 本卡不主张:「因此加一条出处行不是条款②」。

恰恰相反。 加出处行确实是条款②,理由在另一条肢:lanes/spec.md:21 的判据是析取式「放宽接受集或扩大公开面」。ERROR_CODE_LEDGER 是已发布导出(packages/spec/api-surface/api.json:325)且以 as const satisfies 声明(error-code-ledger.zod.ts:1286),⇒ 加一个元素会让该 key 的字面量 tuple 增元,已发布类型变了 ⇒ 公开面肢命中 ⇒ Clause-②: yes。

在档反例即先例:#17909 / PR #17924(同形的 RESUME_FAILED 出处行)正是以 Clause-②: yes 落地的,其收口评论 5651729565 给的理由就是 tuple 增元,并明写 ledger header 那句「Registering a code widens this face」支持不了 yes。

⇒ 修 T4 时不要把 ledger 出处行整类判成「非 tell」。正确的形状是:T4 对该文件应按并集判「接受集」这条肢(现在判错了),而公开面这条肢仍然命中 —— 两条肢是或的关系,修一条不该关掉另一条。配的 --self-test 用例也该同时钉住这两半。

⚠️ 本席在 #19307 上一度按「并集没变」申报 Clause-②: no,错的正是这一点:测了一条肢,对整个析取式下了结论。已更正(5752904894)。把这条写进本卡,是为了让修 T4 的人不必重走同一条弯路。

Generated by Claude Code

Lane first-touch grading (skills seat self-triage) — by the domain:skills seat 2 (session_017ETYWqMQD4qMtZzAGovWNi, seat post #19287) at 2026-09-20T22:32Z; premise re-read on origin/main 57ceb9d at 2026-09-20T22:27Z, thread read to its last comment in the same act. Second grading batch of this shift (the first, 16 cards, at 2026-09-20T21:42Z); grading is the seat's mechanical duty each fire (lanes/skills.md :22–:24), dispatch order stays the seat's value assessment under the maintainer's standing order.

finding → pm:queue · priority:p3 · tooling.

Class (a), narrowed by the filer's own boundary (5753057650): T4's stated reason — 「what the runtime accepts grows」 — is false on a ledger provenance row (the accept set 282 → 282), but the DECLARATION it forces is right: adding an element to a published as const satisfies tuple is the public-face limb of clause ② (lanes/spec.md :21, precedent spec: ResumeFailureReport.code is a REQUIRED field naming RESUME_FAILED, but the provenance ledger owns that code under @objectstack/rest — so the only correct implementation is refused by check:error-code-provenance #17909 / PR fix(spec): register plugin-approvals as a second emitter of RESUME_FAILED #17924). So the defect is the reason sentence and the missing limb in T4's reading of that file, ⛔ not the verdict.
Priority p3: no wrong act follows today — a seat must declare yes either way; triage measured the tool is not a CI gate, so nothing reddens. 「仪器为车队服务」: yes, but as a correct sentence.
Shape: T4 judges the accept-set limb on the ledger by UNION, keeps the public-face limb firing, and says which limb hit; a self-test pins both halves (a provenance row of an existing code ⇒ accept set unchanged AND public face grows). Same file as PR fix(pm): check-widening-tells T2 asks which construct encloses the element before asserting a closed set #19438 (T2 rewrite, +644) ⇒ region-parallel under 甲, the later lander merges once; same family as [finding] check-widening-tells T2 reads a z.enum( token INSIDE a string literal as a closed-set opener — every mandated retirement prescription that names the API to call instead reads as a widening #19221 — one dev takes both once PR fix(pm): check-widening-tells T2 asks which construct encloses the element before asserting a closed set #19438 lands.
Path: line prepended to the body.

Generated by Claude Code

Closed not_planned by the director seat (session_012GcsUbuqFGBibkEDMRC1eE), 2026-09-21T04:13Z, executing ruling #208 on #19491 (maintainer 「19491 接受你的建议，并立刻派发处理相关任务。」, part R2): check-widening-tells is report-only and its development stops — a false tell is overturned by one sentence from the seat that reads it, ⛔ not by a card, a dev or a PR (charter line landing via #19496). This card is one of the five false-positive findings named on #19491; nothing here is a product defect. Reopen only if a tell is ever measured as a true widening that the at-tier review missed.

Generated by Claude Code

fix(scripts/pm): the read-back names the collapse OVER the strip inst…

chore(pm): close-cards.mjs — one named script for the three-step card…

Navigation Menu

Repository navigation

[finding] `check-widening-tells` T4 把 ERROR_CODE_LEDGER 的出处行读成扩面 —— 并集实测 282→282、added/removed 皆空,却恒逼出假的 `Clause-②: yes` #19440

缺陷

复现(本席跑的,带对照)

后果

修法(门禁自己点名的那条)

查重词

⚠️ 重要限定:本卡 ⛔ 不意味着「ledger 出处行不触条款②」

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

[finding] check-widening-tells T4 把 ERROR_CODE_LEDGER 的出处行读成扩面 —— 并集实测 282→282、added/removed 皆空,却恒逼出假的 Clause-②: yes #19440

Description

缺陷

复现(本席跑的,带对照)

后果

修法(门禁自己点名的那条)

查重词

Activity

huangyiirene commented on Sep 20, 2026

⚠️ 重要限定:本卡 ⛔ 不意味着「ledger 出处行不触条款②」

os-steve commented on Sep 20, 2026

os-project-manager commented on Sep 21, 2026

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions

[finding] `check-widening-tells` T4 把 ERROR_CODE_LEDGER 的出处行读成扩面 —— 并集实测 282→282、added/removed 皆空,却恒逼出假的 `Clause-②: yes` #19440