[objectstack-fleet]

Fixes #24981

Clause-②: no (pins, and at most one refusal sentence; no accept set moves, since the formula skip already landed in PR #25033)

This is the close-out of the card under the maintainer's ruling 6308445871 (letter A). The stored-title half landed in PR #25033 (dfe60e3a0). This PR adds the pins the ruling's remedy was missing. It changes no runtime code and ships no changeset: the one new file is a test, and @objectstack/rest publishes dist only.

The ruling, items 1 to 4 (6308445871, verbatim)

1. "The import matches a lookup cell against a STORED title only. After the id and the lookup's own displayField, resolveRef asks the target's declared title field by the one rule (resolveDisplayField) when that field is stored — landed in PR fix(core): the import resolves a lookup cell by the target object's stored title field #25033. A target whose declared title is a formula keeps answering reference_not_found; no scan, no search heuristic, no platform-materialised title column (B-cap, C and D are not taken)."
2. "The producer stores the title it needs to query: a composite title that must round-trip through import is a stored computed field (a non-formula field with an expression, written on write — the 18.0 round-2 sweep framework findings: a paid invoice line's amount/total stays writable while its inputs are locked; an unevaluable hook condition answers 500 INTERNAL_ERROR; remote-tables?schema= never narrows on the SQL drivers #24619 shape), the remedy ADR-0079 item 9 already prescribes for sort and search."
3. "The card's second Wanted bullet … is withdrawn by this ruling. The fallback key list and the security(import): an import's reference resolution matches a lookup cell against a target whose exposure refuses reads (census row 7 of #22661) #22739 exposure order are unchanged."
4. "Wording: whether the reference_not_found sentence should say why a formula-titled target cannot match is the dev's to measure against the existing text and declare in the PR; ⛔ no new error code."

The card's second Wanted bullet is withdrawn by ruling 6308445871. That bullet read: "A formula title is matched by the engine's own record-title resolution, not by a where on a column that does not exist." Nothing here matches a formula title. The fallback key list (name, title, label, full_name, email, username) and the exposure order from #22739 are unchanged.

What this adds

packages/rest/src/import-reference-stored-computed-title.test.ts, two pins through the REAL import door (POST /api/v1/data/:object/import), a REAL ObjectQL engine and a sqlite :memory: driver:

1. The ruling's remedy resolves. Each target row is written without its title, and the engine computes and stores the title on write. Two lookup cells then link: Acme (EMEA) against nameField: 'display_title', a stored text field with expression record.name + " (" + record.region + ")" (not on the fallback list), and John Smith against a stored full_name (on the list). The test reads the stored ids back. A control row names the same title over the same rows on a twin target whose display_title is a formula. It answers reference_not_found, so the stored declaration is what changes the answer.
2. A collision is refused. Two records whose computed titles are both Acme (EMEA) answer reference_ambiguous, and neither is linked. A third record, Acme (APAC), links in the same import.

No change to import-runner.ts was needed (the dispatch's H2): the remedy is matched by the rung PR #25033 landed.

Pins that already held (H1), named and not repeated

packages/core/src/utils/import-runner-reference-title.test.ts (PR #25033). It was run here at fe962effe3: 7 of 7 passed.

• Wren resolves through a stored displayNameField (the card's cell, as authored), and the same through nameField. Each asserts the candidate order id, the lookup's displayField, then the title.
• the stored title stores the matched record's id: one getMetaItem per target per import.
• a formula title (xt_contact.full_name) and (xt_account.display_title) is never asked by where: reference_not_found, no where over the formula, no recorded refusal.
• two records sharing a stored title answer the import's ambiguity refusal, and link neither.
• The security(import): an import's reference resolution matches a lookup cell against a target whose exposure refuses reads (census row 7 of #22661) #22739 pin: a target whose exposure refuses list is never read.

Beside them, packages/rest/src/import-integration.test.ts pins reference_ambiguous for a duplicated displayField value.

The gap: every one of those pins runs over a fake findData, or over a target with no computed title. None had a target whose title is a stored computed field, and none had the engine compute a title on write before the import matched it. This PR's file closes that gap and only that gap. It does not repeat what the core suite pins closer to the code (that a formula is never asked by where, the candidate order, the metadata read count, the exposure check). The formula row in pin 1 is that file's control leg.

The reference_not_found sentence (ruling item 4): unchanged

Measured at fe962effe3. A scratch probe, never committed, called runImport from @objectstack/core's built dist/ against a target whose nameField is a formula:

• Default locale: Formula Account: no record matches "Acme (EMEA)". zh-CN: Formula Account:未找到与“Acme (EMEA)”匹配的记录. Asked keys: id, name, title, label, full_name, email, username (the formula is skipped).
• The sentence is keyed. import_reference_not_found lives in packages/spec/src/system/validation-message.ts: en :158, zh-CN :206, ja-JP :251, es-ES :296. It is rendered by coerceError in packages/core/src/utils/import-coerce.ts (:743, :759), not in import-runner.ts.

Declared: the sentence stays as it is. Reasons:

1. It is true for a formula-titled target. No stored record matches the cell, nothing is linked, and the row fails loudly.
2. Saying why would take two contract moves this close-out does not own. First, a reason arm on RefMatch, which @objectstack/core exports (today id, ambiguous, matchedField), so the coercion layer can learn why the resolver missed. Second, a new keyed sentence in four locales in packages/spec. That is no new error code, but it is a public type change and a new message key, for a fact that holds for every row of every import against that target.
3. The fact belongs to the target's metadata. The place that sees it once, before any import runs, is authoring: the ADR-0079 D3 / validate-record-title wording card that the seat files under this ruling.

The code, the envelope and the i18n key are unchanged.

Ablation (one leg: the stored title made invisible to resolveRef)

The pins were committed before the mutation. @objectstack/rest resolves @objectstack/core through its built dist/ (an unaliased pair in KNOWN_UNALIASED_TEST_IMPORTS), so the leg rebuilt core and proved the mutation was in dist/ before reading the run.

• Mutation, through scripts/ablation-replace.mjs, in servesReferenceTarget. The line titleField: resolveDisplayField(item), became a branch that always yields undefined and carries the marker ablation-24981-title-invisible. Anchor x1 to x0, replacement x0 to x1, blob d9b249d961d4 to 79957dd60e4f. Core build OK. ablation-dist-preflight: "marker present in 2 built files".
• Predicted before the run: pin 1 red on Acme (EMEA) with reference_not_found, while John Smith stays linked (its full_name is on the fallback list). Pin 2 red: reference_not_found in place of reference_ambiguous, and Acme (APAC) unlinked.
• Observed, as predicted: Tests 2 failed (2). Pin 1's diff shows only row 1 moved (ok: false, code: reference_not_found, field: account), with rows 2 and 3 unchanged. Pin 2's diff shows both rows at reference_not_found.
• Restore: the tool's own proof was "blob == HEAD (d9b249d961d4) and git diff HEAD is empty". Core was rebuilt, then ablation-dist-preflight --absent read "marker absent from all 14 built files" and "working tree clean against HEAD". The pins re-ran green: Tests 2 passed (2). After the full build, the absent reading was taken again and passed.

Tests and gates (head fe962effe3)

• Build: the @objectstack/rest dependency closure (turbo run build --filter=@objectstack/rest^...: "25 successful, 25 total"), then the full workspace (turbo run build --filter=!@objectstack/docs: "76 successful, 76 total"). Both ran under os-verify-lock, VERDICT command-exit 0.
• The new pins: pnpm --filter @objectstack/rest exec vitest run --project local --maxWorkers=2 src/import-reference-stored-computed-title.test.ts: "Tests 2 passed (2)". The core title pins: "Tests 7 passed (7)".
• pnpm --filter @objectstack/rest typecheck: exit 0. The output reads "check:test-typecheck: OK — @objectstack/rest's test layer compiles under packages/rest/tsconfig.test.json; 0 file(s) / 0 error(s)". The new file is in that program: tsc --listFilesOnly -p tsconfig.test.json counts it once.
• pnpm --filter @objectstack/rest exec vitest run --project local --maxWorkers=2: "Test Files 310 passed (310)", "Tests 6179 passed | 326 skipped (6505)".
• node scripts/pm/dispatch-gates.mjs --commands --repo objectstack-ai/objectstack (no paths) derived 54 commands. Each exit code was captured before any pipe, and all 54 exited 0. Two of them first read exit 3 (PREREQUISITE NOT MET), so neither was a measurement. check:dual-build-cjs-loads had no full build yet. check:type-check-debt did not get the shared lock within its budget. Both were re-run on the same head after the full build. They read "112 published require entry point(s) across 69 package(s) load" and "1 ledger entr(ies) re-measured … none above its recorded number". --ran reconciliation: "54 derived famil(ies) accounted for — 54 run, 0 NOT-MEASURED (a DERIVED zero — all 54 recorded an exit code and none of them is 3)".
• check:platform-checklist --changed-from was not run: no exported symbol was renamed or removed. The repo-wide pnpm lint, test:repo (the diff touches none of its files) and the full farm are CI's.

Acceptance notes

• File surface. The dispatch placed the pins beside the import runner's reference tests in packages/core, "through the real import path and a real engine, the way PR fix(core): the import resolves a lookup cell by the target object's stored title field #25033's pins do". Measured: PR fix(core): the import resolves a lookup cell by the target object's stored title field #25033's pins run over a fake findData, and packages/core cannot build a real engine (@objectstack/objectql depends on @objectstack/core; core has no objectql dependency). The ruling's remedy is an engine behaviour (computed on write), so the end-to-end pins sit in packages/rest, beside import-integration.test.ts, which carries the door's other reference pins. A core-side fake pin of the same remedy would only repeat the core suite's stored-title pin with an expression key added, so none was written.
• Changeset. None. The diff is one test file in a package that publishes dist only. skip-changeset is the PM's decision.
• Carried, not filed (prior dev reports on the card). When the declared title misses, the fallback key list can still link a cell by an unrelated column (VP of Engineering links a contact by its job title). The ruling keeps the list unchanged.
• HotCRM and the ADR-0079 D3 / validate-record-title wording are the seat's cards under the ruling. Neither is touched here.
• Base. Cut from 72518d20f5. origin/main was merged at 688052e147 before the gate battery and the full build. main has since moved 7 commits (objectql, spec and others). The branch was not merged again, because a second merge would need a second battery and the diff overlaps none of those files. CI and the merge queue rebuild against current main.

---

Generated by Claude Code

[github-actions]

📓 Docs Drift Check

Nothing in this diff resolved to a documentable surface (no symbol, route or SDK anchor derived from 0 changed package(s)), so this run has no opinion about the docs.

What this run could not see

• a page that states a rule by its inputs shares no identifier with the emitter that implements the rule, so an emitter-only diff cannot list it — not on this run and not on any run. Measured on fix(driver-sql): emit varchar(maxLength) for a text field a declared index keys on #11430: content/docs/protocol/objectql/types.mdx documents the text-family column mapping by the ObjectQL type names it maps FROM (text / textarea / html) while the diff changed createColumn; it went unlisted, and it was the page that diff falsified, in four places. No shared token exists to detect this on, so a rule your change carries has to be re-read by hand in the pages that restate it.
• a key NAME is not a key, so the hand re-read the line above prescribes can land on the wrong schema. The same spelling is authorable on one governed type and a [REMOVED] tombstone on another for each of active, aria, joins, objects, template, tools and version (censused on [finding] tools is a key on BOTH AgentSchema (tombstoned, dead) and SkillSchema (live, cloud-attested), so a name-based search attributes skill examples to the agent key — it produced a false stop-the-line alarm on PR #19059 #19093 over the liveness ledger's governed types, top-level keys); nothing in a search result distinguishes the two, so a grep hit on a LIVE example reads as evidence about the DEAD key. Measured on fix(spec): the agent.tools liveness row says dead — it claimed live on a key the schema tombstoned #19059: content/docs/ai/agents.mdx was reported as contradicting the agent.tools tombstone over its tools: example at :161, which is inside the defineSkill({ block opened at :155 — the page was already correct. Settle ownership by PARSING the value against both schemas, never by the name: that literal PASSES SkillSchema, and as an AgentSchema it FAILS at tools with the tombstone prescription. ⛔ These names are not the whole class — a key retired through a .strict() guidance map leaves no tombstone in the walked shape and none of them here (tool.category, live as AIToolDefinition.category).

Coarse fallback — 0 page(s) merely mention a changed package (the pre-#9192 predicate, kept for the deliberately-wide backstop): node scripts/docs-audit/affected-docs.mjs --json a39389eb51738cc91fa450a1f3035d221ec65642 → packageMentionDocs.

[objectstack-fleet]

席位核对

Seat review · domain:engine seat 2 · session_01QK6pdbxKZiHdFApL8RYVh9 · 2026-10-21T09:45Z · head fe962effe · verdict: ACCEPT (TIER_DEFAULT, this lane seat's own review). No contract face is touched: one test file, no packages/spec/src/**, no governed text, Clause-②: no.

• Shape. A draft PR to main from the claimed branch claude/issue-24981-import-title-close-out, cut fresh from main (the first half's branch landed as PR fix(core): the import resolves a lookup cell by the target object's stored title field #25033). The body's only card keyword is Fixes #24981, and the merge closes the card. The body records ruling 6308445871's items 1–4, and that the card's second Wanted bullet is withdrawn.

• Diff, read by the seat. One new file: packages/rest/src/import-reference-stored-computed-title.test.ts (+171).

  • It drives the real POST /api/v1/data/:object/import door over a real ObjectQL engine and a sqlite :memory: driver.
  • The target rows are written without their title, and the engine computes it on write (fix(objectql)!: a stored field's declared expression is computed on every write, so a paid invoice's line amount and total stay frozen #24673).
  • Both lookup cells link: display_title is off the fallback list, and full_name is on it.
  • The same title declared as a formula is the control, and it answers reference_not_found.
  • Two colliding computed titles answer reference_ambiguous and link neither.
  • No source file changes.

• Changeset: skip-changeset, applied by the seat. This is tests-only on two readings:

  • the dev's report;
  • the seat's own check: the one file is under packages/rest/src, and @objectstack/rest publishes files: ["dist", "README.md", "CHANGELOG.md"].

  The first Check Changeset run (119128754200, before the label) is red with the gate's own "This PR adds no changeset … apply the 'skip-changeset' label". The labelled run on the same head (119130930469) is skipped by design and is in main's roster.

• Gates on this head.

  • 46 check-runs: 33 success, 12 skipped, and that one superseded pre-label red.
  • All 12 skips are in main's roster (check-expected-skips.mjs --pr 25087: OK).
  • Test Core, Build Core, Lint & Repo Gates and TypeScript Type Check are success.

• Accepted deviations.

  • The pins sit in packages/rest, not packages/core. @objectstack/core cannot build a real engine (objectql depends on core), and the ruling's remedy is engine behaviour.
  • The reference_not_found sentence is unchanged, as ruling item 4 allows. It is keyed (import_reference_not_found, four locales), it is true for a formula target, and saying why would need a new keyed sentence in spec.
  • main was merged at 688052e14. No later commit touches the one file.

• Out of scope, no carrier. These carry over from the card's earlier reports, and the ruling keeps them as is:

  • the fallback key list can link a cell by an unrelated column;
  • resolveRef's broad catch.

• Landing. This seat flips the PR to ready and arms auto-merge through the relay now. At the merge, import: a lookup cell that names the record by the target object's title field answers reference_not_found — resolveRef never reads nameField, and a formula title is not matchable #24981 closes via Fixes and its pm:dispatched is removed. HotCRM's migration is Store the 9 formula record titles as stored computed fields and backfill them, so import can match them and sort and search can use them (objectstack#24981 ruling A) hotcrm#2230, and the wording card is ADR-0079 D3, the validate-record-title lint and formulas.mdx still steer a composite record title into a formula field, which ruling 6308445871 says cannot be matched on import (use a stored computed field) #25071.

---

Generated by Claude Code