Repository navigation
types(date, datetime): declare dueLike and settledWhen by reference to @objectstack/spec once objectui installs the release carrying objectstack-ai/objectstack#22503 #12297
Description
Filing class: ① product defect, class (b): a contract the protocol declares that objectui's typed authoring face does not yet carry by reference. Filed on objectui#12242's ruling 6137052265 (execution parameters), at PR objectui#12293's merge. Reach: named producers. Three objectstack example apps author dueLike: true on published objects (examples/app-crm activity.object.ts, examples/app-showcase task.object.ts, examples/app-todo task.object.ts).
Restart-when: objectui's installed @objectstack/spec carries objectstack-ai/objectstack#22503 (FieldSchema.dueLike, commit 215e66204)
Restart-touch: pnpm-lock.yaml
- Who acts: the
domain:spec@ objectui seat (objectui#10217), when theRestart-when:line holds. Until then this card ispm:on-holdfrom its filing, on the director's ruling6137052265and the cross-repo unlock rule (installable, not merged). - Scope, this card only:
DateFieldMetadata.dueLikeandDateTimeFieldMetadata.dueLike(packages/types/src/field-types.ts) are typed by reference (SpecField['dueLike']), andsettledWhenis declared beside them by reference (SpecField['settledWhen']). objectuiAGENTS.md#0.1 governs this.DetailViewFieldSchema.dueLikefollows if it still mirrors the field.content/docs/fields/date.mdx,datetime.mdxandcontent/docs/api/schema-reference.mdteach both keys as the spec does.
- Not this card: the renderer's reads. objectui#11815 (
domain:ui,pm:on-holdon the same install event) owns readingdueLikeand the host-evaluatedsettledWhenin the date cells, and retiring the field-name regex (ruling6054882132, letter D). That card can type its reads against this one's declaration, so this card lands first when both restart. - Dedupe: two objectui issue searches, closed included. "dueLike settledWhen deadline date field" returned 2 hits: objectui#11815 (open, the reader half, above) and objectui#11676 (closed, the timeline's overdue bucketing). "field metadata kept positions runtime-only keys hand-off" returned 0 hits; the control ("dueLike settledWhen …" with the known card objectui#11815) hit. Neither hit carries the types declaration.
The reading (this seat, on the day of filing)
- The protocol: objectstack
origin/mainpackages/spec/src/data/field.zod.ts:dueLike: z.boolean().optional()andsettledWhen(a CEL predicate), bothdate/datetimeonly and display only. They landed in objectstack#22503 (215e66204, merged 2026-10-09). - The install face: npm
@objectstack/speclatestis17.7.0(published 2026-10-06, before #22503). objectui's lockfile resolves17.7.0. objectui#11815's triage audit (6087294642) reads215e66204as on the v18 line only. So the restart may coincide with objectui's major bump. - objectui today:
dueLikeis declared locally onDateFieldMetadataandDateTimeFieldMetadata(kept by6137052265), and nothing declaressettledWhen.
Dedupe words: dueLike by reference · settledWhen date field type · deadline semantic objectui types
Generated by Claude Code
Activity
objectstack-fleet commentedon Oct 14, 2026
Unlock scan:
pm:on-hold→pm:queue· the release this card waits on is already installed · triage seat ·session_01AavokzJ5DndAwitDXvKy4U· 2026-10-14T07:38Z215e66204(feat(spec,lint): a date or datetime field declares its deadline: dueLike and settledWhen objectstack#22503,dueLike/settledWhen). That commit is an ancestor of@objectstack/spec@18.0.0-next.0's tag commitd889448b93(git merge-base --is-ancestor, read in this act).mainb9814b334dhas resolved@objectstack/*18.0.0-next.0across the lockfile since PR chore(deps): move @objectstack/* to the next prerelease ^18.0.0-next.0 and follow its contract (objectui#12030) #12337 (objectui#12030).pm:blockedcards, notpm:on-holdcards'Restart-when:lines. It reads both from this round on.6137289497), so date cells: a due-like date on a completed record still renders red "Overdue Nd" (Done tasks flagged overdue) #11815's reads take the declared types.objectstack-fleet commentedon Oct 17, 2026
Triage:
priority:p3→p2·domain:spec→domain:ui·pm:queuestands · triage seat ·session_01AavokzJ5DndAwitDXvKy4U· 2026-10-17T04:56ZTriage seat (objectstack-wide, seat post objectstack-ai/objectstack#6015). ⛔ Not a claim, ⛔ not a dispatch.
domain:ui: objectui'sdomain:specseat is vacant (seat post objectui#10217, "⏳ vacant"), sopackages/typescards route todomain:uiuntil it is filled. The contract-review owed on a types face is unchanged.Generated by Claude Code
objectstack-fleet commentedon Oct 17, 2026
Claim: PM loop round 3 (
priority:p2bug, contract face; filed on objectui#12242's ruling6137052265, unlocked by the triage seat's install scan6163379426, re-graded p2 and routed todomain:uibecause objectui'sdomain:specseat is vacant,6229884299; the blocker of objectui#11815 and objectui#12726)Session:
session_01SJn1GEkrbbBQkhGC26wEDwAccount:
os-sales(the seat's linked user asGET /useranswers it; the card's assignee)Branch:
claude/issue-12297-duelike-by-referenceWorktree:
objectui-issue-12297(in the cloud session's own container)Domain:
domain:uiSeat:
domain:ui#4File surface:
packages/types/src/field-types.ts(DateFieldMetadata.dueLikeandDateTimeFieldMetadata.dueLiketyped by reference to@objectstack/spec,settledWhendeclared beside them by reference);DetailViewFieldSchema.dueLikeand any zod mirror of these members underpackages/types/src/zod/only if they still mirror the field;content/docs/fields/date.mdx,content/docs/fields/datetime.mdx,content/docs/api/schema-reference.md; tests beside; one.changeset/12297-*.md(@object-ui/types). ⛔ The renderer's reads (objectui#11815) andpackages/components/src/ui/**stay out. Stop on breach; explain in the report.Container & model:
S,mode:cloud,model: opus(cloud on the maintainer's stance 「可以的话，ui 后续开发尽量用云 session，并且要求截图给我看」; lane scope on 「你也可以接ui车道的卡」; batch 3 on 「当前任务处理完，后续并发降到3」, this session's chat, 2026-10-16)Clause-②: yes
Clause-② reason:
@object-ui/typespublishesDateFieldMetadataandDateTimeFieldMetadata;dueLikeis re-typed by reference andsettledWhenis a new optional member. The seat'sTIER_DEFAULTreview comes before ready.Responsibility: n/a — a protocol key objectui's typed authoring face does not carry by reference
Thread-read: 6229884299
Serial constraints cleared: no open objectui PR touches
packages/types/src/field-types.tsor the three docs pages (read at dispatch). objectui#11815 (the reader half) stays on hold and types its reads against this card's declaration after it lands. Not a governed surface (check-governed-queue-guard.mjs --teston the paths: none matched).objectstack-fleet commentedon Oct 17, 2026
os-dev-report
{ "issue": 12297, "status": "done", "branch": "claude/issue-12297-duelike-by-reference", "pr": 12861, "session": "session_01Qe458QJ3QyHSzgf9t1dDEH", "premise_still_valid": true, "summary": "`DateFieldMetadata` and `DateTimeFieldMetadata` now take `dueLike` from the spec as `SpecField['dueLike']`, using the file's own `Field as SpecField` idiom (`Field` = `z.input<typeof FieldSchema>`). Beside it, on both interfaces, I declared `settledWhen?: SpecField['settledWhen']`, with doc comments taken from the spec: a CEL predicate, display only, `date`/`datetime` only, `dueLike: true` required. The comments also say two things the runtime does not do yet: today's cells still fall back to the field name when `dueLike` is absent, and `settledWhen` is declared but not rendered (both are objectui#11815's job). The docs pages date.mdx, datetime.mdx and schema-reference.md teach both keys the same way, and one minor changeset covers @object-ui/types. `DetailViewFieldSchema.dueLike` and its zod mirror are left untouched, for the reason in open_questions. No zod mirror of field metadata exists, so nothing was added there. All three PM mechanism assumptions held when measured on main `024d446`.", "tests": "At HEAD 868d7b8. `pnpm --filter @object-ui/types test` (run once, foreground, under os-verify-lock): VERDICT command-exit 0, Test Files 386 passed (386), Tests 10513 passed (10513). Targeted repo-root run `pnpm exec vitest run` on field-metadata-deadline-by-reference-12297.test.ts and field-metadata-family-retired-12242.test.ts: 2 files, 27 tests passed. The new pin, packages/types/src/__tests__/field-metadata-deadline-by-reference-12297.test.ts, has three parts. (1) Type half: exact `Equal` checks that `DateFieldMetadata` and `DateTimeFieldMetadata` `['dueLike']` and `['settledWhen']` equal `SpecField[KEY]`. (2) Control: `'dueLike'` and `'settledWhen'` are not in `keyof` `TimeFieldMetadata` or `TextFieldMetadata`, and fresh `time`/`text` literals carrying them are refused via `@ts-expect-error`. (3) Runtime half against the installed `FieldSchema`: both keys parse on date (bare string) and datetime (envelope); `dueLike` on `time` is refused at path `dueLike`, with a minimal `time` field as the control; `settledWhen` without `dueLike: true` is refused at `settledWhen`, and the same field with it passes. Ablation, once, after committing the fix: I replaced `DateFieldMetadata`'s `settledWhen?: SpecField['settledWhen']` with a local `settledWhen?: string`. On disk: marker grep count 1, and exactly 1 by-reference line left (the datetime one). `tsc -p tsconfig.test.json` then exited 2 with `field-metadata-deadline-by-reference-12297.test.ts(44,10): error TS2344: Type 'false' does not satisfy the constraint 'true'.`, which is exactly the Date settledWhen Equal row. I restored with `git checkout HEAD --` under an EXIT trap. The restored blob 81b45d441d7bd775107cdd243fe3453cc896eadd matches the HEAD blob, and `git diff HEAD` is empty. The test imports '../field-types.js' relatively, so tsc reads src and no dist rebuild was needed for the ablation.", "gates": "Local, at 868d7b8 unless stated; exit codes captured before any pipe. `pnpm --filter @object-ui/types type-check` EXIT=0, script name echoed; it runs tsc, tsconfig.examples and tsconfig.test, and tsconfig.test includes the new file. `pnpm --filter @object-ui/types lint` EXIT=0 with 0 errors; the 3 warnings on field-types.ts are pre-existing (no-explicit-any; the base blob gives the same 3). `pnpm --filter @object-ui/types build` via lock: VERDICT command-exit 0, dist completeness verified, and dist/field-types.d.ts carries settledWhen. Doc-snippet build closure (`turbo run build` with the `--build-filter` from check-doc-snippet-types, concurrency 2) via lock: VERDICT command-exit 0, 35/35 tasks. `pnpm check:doc-snippets` EXIT=0 (801 of 801 blocks judged, 0 failed). `pnpm check:doc-examples` EXIT=0. `node scripts/check-doc-json-fences.mjs` EXIT=0. `node scripts/check-doc-links.mjs` EXIT=0. `node scripts/check-doc-component-types.mjs` EXIT=0. `node scripts/check-doc-example-ids.mjs` EXIT=0. `node scripts/check-control-bytes.mjs` EXIT=0. `pnpm check:spec-symbols` EXIT=0, with output identical before and after the change (diff of the ✅ lines is empty), and the member-citation leg reads 'nothing cites a key its spec symbol does not declare'. `node scripts/check-changeset-presence.mjs` EXIT=0 (1 changeset declared for 3 released source files). `pnpm check:new-line-citations` EXIT=0 (0 new citations). Control-byte grep on the changed files: no hits. Model-name grep over `git log --format=%B 024d446..HEAD`: no hits. The first run of check-doc-snippet-types and check-doc-example-types exited 2 with PREREQUISITE (closure unbuilt), not as a measurement; both are green after the closure build. objectui has no dispatch-gates.mjs, so no `--ran` reconciliation is possible; I hand-derived this list from repo-gates.yml. CI check-runs: NOT MEASURED, because no PR exists yet (the PM opens it).", "mcp_calls": "6 in total. claude-code-remote add_repo x1 (objectstack-ai/objectstack, access read; it answered read_available and attached nothing). github issue_read x4: get #12297, get_comments #12297, get #11815, get_comments #12242 (for ruling 6137052265). ToolSearch x1 to load issue_read (not an MCP call itself). No write tool was used.", "api_writes": "0 REST writes. Two git pushes to objectui branch claude/issue-12297-duelike-by-reference: the empty branch (write probe), then the commits 8d1a432 and 868d7b8. No PR, no comment, no label write.", "open_questions": [ { "question": "DetailViewFieldSchema.dueLike (packages/types/src/views.ts and the zod/views.zod.ts mirror): is it re-typed by reference? Ruling: it follows only if it still mirrors the field.", "options": [ "A: leave it untouched (taken)", "B: re-type it as SpecField['dueLike'] and add settledWhen to it" ], "recommendation": "A. What I found: the installed spec declares `dueLike` only on `FieldSchema` (a grep of `@objectstack/spec@18.0.0-next.2` src finds it only in data/field.zod.ts), so no spec view schema carries it. The interface's own doc calls it 'a different key of the same name' on the authored detail-view entry. Its runtime meaning also differs from the spec's: an absent key falls back to the name convention, and the authored entry wins over the object field. Its type is already `boolean | undefined`, equal to the spec's. Against the four axes: B would claim spec provenance for a key the spec does not declare on any view (axis 3: declaring what the contract does not say). Adding `settledWhen` there would create a new capability with no spec row and no reader (axis 4). Neither has any business pull today (axis 1). Whether the detail entry keeps `dueLike` after objectui#11815 retires the name fallback belongs to that card." }, { "question": "Where my contract (objectstack .claude/agents/os-dev.md at 265ef64) and this dispatch disagree", "options": [ "Followed the dispatch on writes; contract notes follow" ], "recommendation": "(1) The contract makes opening the draft PR and posting the os-dev-report comment the dev's job. The dispatch says not to, because the fleet-write relay is refused from this container. I did neither, put `pr_title_draft`/`pr_body_draft` in this report, and did not route the writes through MCP. The contract itself sends every GitHub write through the relay and never through MCP, so I read this as the channel being down rather than a real conflict; the PM posts this report on the card. (2) The harness attribution reminder asked for a model-named Co-Authored-By trailer. I used AGENTS.md's model-free pair instead, as both AGENTS.md and the contract direct. (3) The contract's `scripts/ablation-replace.mjs` and `ablation-dist-preflight.mjs` live only in objectstack, not in objectui. The ablation used an inline replace that asserts exactly one match, grep counts on both sides, and HEAD-blob hash restore. No dist path was involved because the test compiles src. (4) The worktree is not removed: the contract removes it after the PR opens, and I did not open the PR. (5) objectui AGENTS.md names the lock at /home/user/objectstack/scripts/pm/os-verify-lock.sh, but in this container the read-only clone is at /home/user/objectstack-ai/objectstack. I used that path." }, { "question": "Doc pages: how to teach `dueLike` when the spec says 'absent = not a deadline, nothing inferred from the name' but today's cells still infer from the name", "options": [ "A: state both, the spec meaning and today's fallback, with the retirement owed by objectui#11815 (taken)", "B: teach only the spec meaning" ], "recommendation": "A. B would have the docs promise behaviour the runtime does not deliver: an author writing `dueLike: false` on `due_at` would still see red. The ruling bans documenting undelivered capability (axis 3), and spec > implementation > docs means the gap is objectui#11815's to close, not the docs'. The pages therefore say which of the two answers is the spec's and which is today's renderer." } ], "out_of_scope_findings": [ "carrier: objectui#11815 · noted, not filed: the spec reads an absent or false `dueLike` as 'not a deadline, nothing inferred from the name', but `resolveDueLike` (@object-ui/fields) still falls back to the field-name regex and ignores `false`. The new comments and docs state this gap, and retiring the regex is already that card's scope." ], "pr_title_draft": "fix(types): type dueLike by reference and declare settledWhen on date and datetime field metadata", "pr_body_draft": "(elided here: it is the body of objectstack-ai/objectui#12861, the dev's draft verbatim)" }Relayed by seat
domain:ui#4(session_01SJn1GEkrbbBQkhGC26wEDw) from the cloud session's final report (degraded channel: the session pushed and returnedpr: null). The seat opened #12861 fromclaude/issue-12297-duelike-by-referenceat868d7b8and setprabove. The three open questions (DetailViewFieldSchema.dueLike, the contract notes, how the docs teach the name fallback) are ruled in the ACCEPT on the PR, and the seat'sTIER_DEFAULTcontract review follows it.Generated by Claude Code
objectstack-fleet commentedon Oct 17, 2026
Landed · PR #12861 merged as
8106db1onmain· read 2026-10-17T05:53Zorigin/main8106db1DateFieldMetadatainpackages/types/src/field-types.tsdueLike?: SpecField['dueLike'](:600),settledWhen?: SpecField['settledWhen'](:615)DateTimeFieldMetadatadueLike(:646),settledWhen(:654), bothSpecField[...]DetailViewFieldSchema.dueLikefield-metadata-deadline-by-reference-12297.test.tsEqualrows, a keyof CONTROL, a runtime half.changeset/12297-date-deadline-keys-by-reference.md@object-ui/typesminorState change: the card closed as completed through
Fixes #12297, andpm:dispatchedis removed in this act. The cloud sessionsession_01Qe458QJ3QyHSzgf9t1dDEHis archived.Records of this card:
6230158528;6230165399;TIER_DEFAULTcontract review:6230173962(PASS);6230267513.Unblocked: objectui#11815 (the reader half: reading
dueLike/settledWhenin the date cells and retiring the name regex) can restart against this declaration, and objectui#12726 waits on it.Generated by Claude Code