[plugin-calendar] declare the remaining spec keys `object-calendar` reads — `data` · `staticData` · `loading` (objectui#8201 slice 2b)

[os-justin]

Slice 2b of objectui#8201, split by package under the PM ruling Q2 = B on that card. Filed by the domain:ui PM seat. ⛔ Not claimed.

Why split by package

Each member pin has to be measured at its own sink, and objectui#8223 proved how easily a sink is assumed wrong — its sort pin could never have been an identity pin because convertSortToQueryParams builds a new map, while objectui#8212's landed rider note had predicted the opposite. Splitting also keeps each PR inside the plugin package's own __tests__ dir, where the worked precedents live.

Scope — three keys

object-calendar (both tags: plugin-calendar:object-calendar and view:calendar):

key	declared arm	note
data	array	owes a MEMBER_PINS entry
staticData	array	owes a MEMBER_PINS entry
loading	boolean	⚠️ scalar — see below

All three were measured genuinely spec-declared in slice 1, with a control key drawing unrecognized_keys on the same safeParse call shape while none of these was refused by name. ⛔ objectui#8172's limit trap does not apply.

⚠️ Why loading is here and not in slice 1

loading is scalar, so it owes no member pin and slice 1's boundary (scalar-armed keys only) would otherwise have taken it. It was deliberately held back: it is inert without data — a loading flag on a block that cannot be handed rows says nothing — so it ships with the key that makes it meaningful.

⇒ if your measurement contradicts that (e.g. loading is honoured independently of data), say so and declare it on its own terms; the pairing is slice 1's reasoning, not a measured requirement.

⚠️ The three-part obligation each array-armed declaration owes

Since objectui#8212 landed:

1. the inputs entry;
2. deletion of its UNPUBLISHED_EXEMPTIONS entry in the same change;
3. a MEMBER_PINS entry — ⛔ never a member-pin exemption. MEMBER_PIN_EXEMPTION_CEILING stays 62; an assertion pins four keys by name so a fifth cannot ride that headroom.

⚠️ Both halves of the shrink-only backlog ceiling move (toBeLessThanOrEqual and toBe, same it()) — moving only toBe banks headroom the ratchet exists to refuse.

⚠️ Measure each sink

Three worked precedents on main, two of them for this very package:

• packages/plugin-calendar/src/__tests__/ObjectCalendar.filterIsNotAConfigSlot-7711.test.tsx — identity (toBe)
• packages/plugin-calendar/src/__tests__/ObjectCalendar.sortMembersReachTheWire-8171.test.tsx — not identity; asserts what is read inside the member
• packages/plugin-kanban/src/__tests__/ObjectKanban.filterMembersReachTheWire-8176.test.tsx

⛔ Do not assume which shape applies. ⚠️ objectui#8269 showed a sharper version of the same trap: a resolver answering which field and one answering which column the rows carry coincide only when no alias is written.

⚠️ One measured oddity in this package worth knowing before writing the data pin: getCalendarConfig (packages/plugin-calendar/src/ObjectCalendar.tsx:156) returns the block as soon as schema.calendar is truthy, so calendar: {} yields startDateField: undefined and every record lands in unscheduledRecords — degraded-but-visible, which is the ruled objectui#7071 behaviour rather than a defect. Recorded because the spec now refuses calendar: {} at its own doors while objectui's .partial() mirror accepts it, so the two faces disagree by design.

Related

objectui#8201 (parent, slice 1 = PR #8309) · objectui#8212 · objectui#8223 · objectui#7071 · objectui#8170 (the refusal screen on this renderer still names titleField, which spec 17.3.0 made optional).

[os-justin]

Claim: session session_01YBWFb5YgMU5dw8p2VKj16S · branch claude/issue-8314-calendar-data-staticdata-loading

PM dispatch (domain:ui seat). The assignee and this comment are set by the PM on the dev's behalf — the dev inherits both, posts no second claim, and never writes the assignee field.

Sequencing note for the reader: slice 1 (PR #8309) and the board half of slice 2 (PR #8328, object-kanban's four array/object-armed keys) are both on main / in the queue ahead of this card, so the backlog ceiling this branch starts from must be re-derived on its own base, not taken from this card's text. After #8328 the backlog is 4 — one object-kanban key (quickAdd, carved out separately under objectui#8285) and three object-calendar keys, which are exactly this card's three.

---

Generated by Claude Code

[os-justin]

os-dev-report

{
  "issue": 8314,
  "status": "done",
  "branch": "claude/issue-8314-calendar-data-staticdata-loading",
  "pr": "https://github.com/objectstack-ai/objectui/pull/8340",
  "premise_still_valid": true,
  "summary": "Declared `data`, `staticData` and `loading` on both `plugin-calendar` registrations (one shared OBJECT_CALENDAR_INPUTS list, so the two tags cannot drift), deleted the three matching UNPUBLISHED_EXEMPTIONS entries in the same change, and registered a MEMBER_PINS entry for each of the two array-armed keys — never an exemption, so MEMBER_PIN_EXEMPTION_CEILING stays 62. Both halves of the shrink-only backlog ceiling moved. Each key was measured at its own sink and the sinks differ: `data` and `loading` are read off the PROPS channel at the renderer boundary (resolveExternalData / resolveExternalLoading in index.tsx), `staticData` off the SCHEMA by the shared record-source ladder's rung 2. One rider was mechanically required and is in the file that owns it: `staticData` joins SUPERSEDES_BINDING in apps/console/src/__tests__/public-block-binding-reach.test.tsx, with its ladder rung quoted, because that probe fills every declared input and then asserts the block queried its objectName — and staticData is one rung ABOVE objectName, so filling it is the author telling the block not to fetch. The PM inherited claim on #8314 names this session and this branch; no second claim was posted and the assignee field was never written.",
  "backlog_ceiling": {
    "started_from": 8,
    "how_read": "Read off my own base, not the brief: `git worktree add --no-track ... origin/main` put BASE at d8ab2dc0c, where apps/console/src/__tests__/registry-inputs-spec-parity.test.ts's `the objectui#8176 backlog has a ceiling` it() read `toBeLessThanOrEqual(8)` and `toBe(8)`, and UNPUBLISHED_EXEMPTIONS carried five object-kanban keys plus three object-calendar keys on LAZY_REGISTERED_BLOCKS. The brief predicted 4 on the assumption that #8328 was already on main; it was not.",
    "first_commit_moved": "8 -> 5 (both halves, same it())",
    "then": "objectui#8313 (PR #8328) merged while this branch was open; main moved d8ab2dc0c -> 614d85b19 and the merge CONFLICTED on exactly those lines plus the exemption entries.",
    "final": "1 — re-derived on the merged tree rather than by picking a side. The whole remaining backlog is `object-kanban.quickAdd`, the escalated key (objectui#8285), untouched here.",
    "member_pin_exemption_ceiling": "62 before, 62 after — unchanged. MEMBER_PIN_EXEMPTIONS length unchanged (the four objectui#8176 names are still the only entries on lazy-registered blocks). My two keys entered the member population and were answered with PINS."
  },
  "member_population_readings": {
    "member_pins_registered_by_this_change": [
      "object-calendar.data -> packages/plugin-calendar/src/__tests__/ObjectCalendar.recordSourceMembers-8314.test.tsx",
      "object-calendar.staticData -> the same file (it names both keys and the block, which is what the locator requires)"
    ],
    "member_pin_exemptions_added": "none",
    "of_declared": "none, and derived: both spec rows are z.array(z.unknown()), so string/number/boolean/array/object ALL parse at the member position. objectui#8067's rule declares `of` only for a single-kind member contract, so no `of` is available and the READ SITE is the whole member contract there is — the same reading object-calendar.filter and .sort already carry."
  },
  "per_key": [
    {
      "key": "data",
      "declared_arm": "array",
      "spec_row": "z.array(z.unknown()).optional() — 'Pre-fetched records — skips the internal fetch'",
      "read_site": "packages/plugin-calendar/src/index.tsx — resolveExternalData(rest.data) on the PROPS channel, forwarded as the component's `data` prop, where ObjectCalendar's Array.isArray(externalData) makes hasExternalData and seeds the row state.",
      "member_pin_question": "Which keys are read INSIDE a member? A member is a RECORD: the fields the declared `calendar` config names (startDateField / endDateField / titleField) plus `id`. Rows: every member arrives in authored order with its title and id; `allDay` is derived PER MEMBER from the end field (two members differing only there get opposite verdicts); a member with no start value is counted in the unscheduled area rather than dropped or given a fabricated date; a member with no `id` gets a synthesised `event-N` rather than being discarded.",
      "ablation": "Deleted `data={externalData}` from the boundary's JSX. Mutation proved on disk (git hash-object ac9230f3 -> 184535e4, anchor occurrences HEAD=1 disk=0). Result: 7 failed / 4 passed — every rows-are-drawn row red. Restored by state: hash back to ac9230f3, `git diff HEAD` empty."
    },
    {
      "key": "staticData",
      "declared_arm": "array",
      "spec_row": "z.array(z.unknown()).optional() — 'Static inline records'",
      "read_site": "packages/core/src/utils/record-source.ts — resolveRecordSourceConfig rung 2, reached from ObjectCalendar's dataConfig memo off schema.staticData; wrapped as { provider: 'value', items } and drawn by the fetch effect's inline branch. NOT read on the props channel at all.",
      "member_pin_question": "The same member question as `data` (identical record keys, identical unscheduled treatment), PLUS the two POSITION claims its description makes and no other direction of the gate can see: rung 2 is BELOW `data` (an authored data wins and staticData contributes nothing) and ABOVE `objectName` (a calendar carrying both draws the inline rows and issues no query).",
      "ablation": "Deleted the rung-2 block from packages/core/src/utils/record-source.ts. Mutation proved on disk (ca4e209a -> 64ec582a, anchor HEAD=1 disk=0). Result: 3 failed / 8 passed — exactly the three staticData-dependent rows; `data wins over staticData` correctly stayed green. Restored by state (hash equal, git diff HEAD empty)."
    },
    {
      "key": "loading",
      "declared_arm": "boolean (scalar — owes no member pin)",
      "spec_row": "z.boolean().optional() — 'External loading state (honoured only alongside `data`)'; the string spelling is refused (invalid_type, expected boolean)",
      "read_site": "packages/plugin-calendar/src/index.tsx — resolveExternalLoading(rest.loading) on the PROPS channel; ObjectCalendar applies it only under hasExternalData (initial useState and the sync effect both gate on it).",
      "member_pin_question": "n/a — scalar. Its DESCRIPTION makes a position claim instead, and that claim is pinned behaviourally: `loading: true` alongside an array `data` replaces the grid with the 'Loading calendar...' placeholder, and the IDENTICAL flag on a `staticData`-fed calendar changes nothing and the rows are drawn. One controlled pair, only the row-carrier differs, opposite verdicts.",
      "ablation": "Deleted `loading={externalLoading}` from the boundary's JSX. Mutation proved on disk (ac9230f3 -> 5d990d36, anchor HEAD=1 disk=0). Result: 1 failed / 10 passed — exactly the loading-honoured row. Restored by state."
    }
  ],
  "clause_2_control": "MEASURED on the installed @objectstack/spec 17.3.0, on the SAME strict safeParse call shape, not asserted: safeParse({objectName:'visit', data:[{id:1}], staticData:[{id:2}], loading:true}) -> success=true (no key refused by name), while safeParse({objectName:'visit', bogusProbeKey:'x'}) -> success=false with [{\"code\":\"unrecognized_keys\",\"path\":[],\"keys\":[\"bogusProbeKey\"]}]. So the probe draws unrecognized_keys on the same call while none of the three does. objectui#8172's `limit` trap does not apply here. Pinned as a row in packages/plugin-calendar/src/__tests__/dataKeysAreDeclaredAndHonoured-8314.test.ts.",
  "tests": "All verdicts quoted from the runner's own summary, re-run after merging origin/main (614d85b19) into the branch. FINAL TREE: `pnpm exec vitest run apps/console/` -> 'Test Files  89 passed (89) / Tests  1069 passed (1069)', EXIT=0. `pnpm exec vitest run packages/plugin-calendar/ packages/plugin-kanban/` -> 'Test Files  60 passed (60) / Tests  420 passed (420)', EXIT=0. `pnpm --filter @object-ui/plugin-calendar type-check` EXIT=0 (script echoed: tsc --noEmit && tsc -p tsconfig.test.json — so the new test files ARE type-checked). `pnpm --filter @object-ui/console type-check` EXIT=0 (tsc --noEmit && tsc -b tsconfig.node.json --force). Both ran only AFTER `pnpm --filter '@object-ui/console^...' build` (and `'@object-ui/plugin-calendar^...' build` earlier) — a stale dist reports TS2307/TS2882 across untouched files and is not a verdict. Lint is package-scoped: plugin-calendar EXIT=0 'x 219 problems (0 errors, 219 warnings)', console EXIT=0 'x 211 problems (0 errors, 211 warnings)'. Gates: check-control-bytes 'OK (scanned 6636 tracked text file(s); skipped 85 binary)'; check-changeset-presence '5 source file(s) of 2 released package(s) changed, and this change declares 1 changeset(s)'; check-changeset-no-major 'No changeset declares a major bump'; check-sdui-registration-pins 'All 16 registration(s) a sideEffects array promises are present in the built console (518 chunks weighed; the 3 ruled control(s) are in the derived set)' — this one first returned EXIT=2 PREREQUISITE NOT MET and was re-run after `pnpm --filter @object-ui/console build` rather than recorded as a failure; check-element-data-source-declaration 'OK - 13 gate-consuming file(s) checked'; check-side-effects-array, check-vi-mock-specifiers, check-vi-mock-inherit, check-unreferenced-sources, check-spec-symbols all OK. NOT MEASURED, stated rather than dropped: `pnpm run check:changeset-no-major` is not a root script name (ERR_PNPM_NO_SCRIPT) — re-run as `node scripts/check-changeset-no-major.mjs`, which is the real gate and is green; the first result was a script-name miss, not a red gate. ABLATIONS — six legs, every one from the COMMITTED implementation, mutating a READ SITE (never a pin), each proving the mutation reached disk by `git hash-object` differing from `git rev-parse HEAD:PATH` plus an anchor-occurrence count that changed, and each restored BY STATE (`git checkout HEAD -- ABSOLUTE_PATH`, then hash equal to the HEAD blob AND `git diff HEAD` empty; `git status --short` empty at the end). No build stands between a source mutation and the measurement: vitest.config.mts's resolve.alias maps every @object-ui/* specifier to that package's src, verified by reading the alias table. (1) delete data={externalData} -> 7 failed / 4 passed. (2) delete the ladder's staticData rung -> 3 failed / 8 passed. (3) delete loading={externalLoading} -> 1 failed / 10 passed. (4) re-add the deleted object-calendar.data UNPUBLISHED_EXEMPTIONS entry -> 2 failed: 'carries no stale unpublished-key exemption' AND 'the objectui#8176 backlog has a ceiling'. (5) delete the object-calendar.staticData MEMBER_PINS entry -> 2 failed: the member-shape census and 'object-calendar pins the member shape of every array/object input it declares'. (6) strip loading's POSITION sentence from its description -> 2 failed, the description row on BOTH tags, with the message naming the missing phrase.",
  "mcp_calls": "11 — 2 issue_read (get + get_comments on #8314), 3 search_issues (the first two returned 'API rate limit already exceeded', the third succeeded and is the dedup read behind the new issue), 1 create_pull_request, 1 update_pull_request, 2 pull_request_read (body read-back after create and after edit), 1 issue_write (creating #8348), 1 add_issue_comment (this comment). Note for the PM: no REST channel exists in this container — an unauthenticated GET to api.github.com returns 403 'GitHub access is not enabled for this session', and `gh` is not installed — so MCP was the only write channel and the only search channel available.",
  "open_questions": [],
  "out_of_scope_findings": [
    "filed as #8348 (label `finding`, no assignee): `object-calendar` honours a SECOND, off-spec spelling of `data` — an authored `{ provider: 'value', items: [...] }` reaches the shared record-source ladder's rung 1, is returned verbatim as the config, and the rows are drawn with no query issued (measured through the real SchemaRenderer: grid present, both titles rendered, find calls 0), while ComponentPropsMap refuses that shape by kind (invalid_type, expected 'array'). That is the AGENTS.md #0.1 second-dialect shape. Deduped with a targeted MCP search_issues before filing: 21 hits, nearest is the CLOSED #7313 (same ladder, seen from the objectName-is-REQUIRED side), no duplicate. Resolution is a product/contract ruling (narrow the renderer vs widen the spec) and touches five renderers sharing the ladder, so it was escalated rather than acted on."
  ],
  "contradictions_with_the_brief": [
    "THE BACKLOG NUMBER. The brief said 'After #8328 lands the backlog should be 4'. On my base it was 8 — PR #8328 was NOT on main when this branch was cut (main was d8ab2dc0c; #8328 landed later as 8a4a1064a). I trusted my base, as instructed, and the first commit moved 8 -> 5. #8328 then merged mid-flight and the merge conflicted on those exact lines; resolving it re-derived the number to 1. Anyone reading the brief's '4 -> 1' will find neither number in the diff.",
    "THE `data` DOUBLE-GUARD IS REAL HERE BUT ITS SHAPE IS DIFFERENT, and the difference matters. On object-kanban the brief reports that removing either guard alone left every row green. On object-calendar the guards are not redundant for the ROWS: deleting the boundary forward turns every rows-are-drawn row red. What IS unfailable here is the 'no query was issued' half — because a bare array under schema.data reaches the ladder's rung 1, is returned as a config carrying no `provider`, and therefore matches no fetch branch, so the query stays unissued on the broken tree too. MEASURED: under ablation (1) the line `expect(dataSource.find).not.toHaveBeenCalled()` PASSED and the very next line failed. The pin was written around this: the query row is ordered query-first, labelled a companion, and the load-bearing claim is the rows. Porting kanban's conclusion unchanged would have produced the wrong pin.",
    "THE BRIEF'S SCOPE TABLE WAS INCOMPLETE IN ONE MECHANICAL RESPECT. Declaring `staticData` immediately reddened apps/console/src/__tests__/public-block-binding-reach.test.tsx ('object-calendar declares an objectName input but made no data call naming probe_object__c'). That probe fills EVERY declared input, and staticData is one ladder rung ABOVE objectName. The file already owns the ledger for exactly this (SUPERSEDES_BINDING, which already contained `data` and `customFields`), so the fix was one entry with the guard quoted, as that ledger's own docblock requires. A dispatch that only listed the parity gate would have left this red.",
    "`loading`'s COUPLING TO `data` HELD UNDER MEASUREMENT — reported because the brief explicitly invited the opposite. `loading: true` + array `data` shows the loading placeholder and hides the grid; the identical `loading: true` + `staticData` draws the rows and shows no placeholder. Slice 1's reasoning was correct, so `loading` ships with `data` and its description names the coupling.",
    "NO SHARED VERIFICATION LOCK EXISTS IN objectui. The dispatch-level resource discipline names `scripts/pm/os-verify-lock.sh`; this repo's scripts/pm/ contains only check-half-states.mjs. Runs were therefore direct, kept targeted (per-package filters, no repo-wide `pnpm test`), and NODE_OPTIONS=--max-old-space-size=4096 was set on every heavy command."
  ],
  "notes_for_the_pm": [
    "PR #8340 is a DRAFT with `Fixes #8314` on its first line; it is NOT governed (`node scripts/check-governed-queue-guard.mjs --test` on all four paths: 'NOT GOVERNED - 4 path(s) checked against 5 governed surface(s); none matched'), so the ordinary review-and-queue route applies and no maintainer-brief section was added.",
    "A changeset is present (.changeset/8314-calendar-record-source-authoring-keys.md, `@object-ui/plugin-calendar: patch`, matching slice 1's precedent), so the `skip-changeset` label is NOT applicable and was not applied. Labels on the PR are the labeler's own: plugin, apps, tests.",
    "CI had not converged when this report was filed; PR mergeable_state read `blocked` (checks pending on a draft), not `dirty`. The branch head is 86e485633.",
    "The README was deliberately not touched: slice 1 (PR #8309) set the precedent for this declaration class, and packages/plugin-calendar/README.md's Schema API section documents the `CalendarView` component rather than `object-calendar`'s authoring keys."
  ]
}

---

Generated by Claude Code