feat(spec,driver-sql,cli): one physical representation for the NUMERIC column family, read by all three producers - #16887

📓 Docs Drift Check

This PR changes 4 package(s): @objectstack/cli, @objectstack/driver-sql, @objectstack/service-analytics, @objectstack/spec, touching 21 documentable anchor(s). ⚠️ 5 changed file(s) yielded no anchor (packages/services/service-analytics/src/measure-result-type.ts, packages/spec/api-surface/data.json, packages/spec/export-origins/data.json, …), so the pages documenting them are NOT COVERED by this run — this is not a clean bill of health for those files.

10 hand-written doc(s) NAME something this change touched and may need an implementation-accuracy re-verification:

• content/docs/api/data-flow.mdx (via os generate (command, read off packages/cli/src/commands/generate.ts))
• content/docs/data-modeling/drivers.mdx (via SqlDriver (symbol, a top-level class))
• content/docs/data-modeling/index.mdx (via SqlDriver (symbol, a top-level class))
• content/docs/deployment/cli.mdx (via os generate (command, read off packages/cli/src/commands/generate.ts))
• content/docs/permissions/tenant-audit-census.mdx (via SqlDriver (symbol, a top-level class))
• content/docs/plugins/packages.mdx (via SqlDriver (symbol, a top-level class))
• content/docs/protocol/kernel/index.mdx (via SqlDriver (symbol, a top-level class))
• content/docs/protocol/kernel/lifecycle.mdx (via SqlDriver (symbol, a top-level class), os generate (command, read off packages/cli/src/commands/generate.ts))
• content/docs/protocol/objectql/query-syntax.mdx (via SqlDriver (symbol, a top-level class))
• content/docs/protocol/objectql/types.mdx (via NUMERIC_COLUMN_REPRESENTATION (symbol, a top-level const object), SqlDriver (symbol, a top-level class), os generate (command, read off packages/cli/src/commands/generate.ts))

⛔ 1 release-owned page(s) also name something this change touched. These are read-only:

• content/docs/releases/v17.mdx (via SqlDriver (symbol, a top-level class))

content/docs/releases/ is RELEASE-OWNED (AGENTS.md "Documentation Guardrails"): release
notes are written centrally at release time, and a code PR that edits them is the exact PR
that guardrail exists to stop. They are still audited — read-only. If one of them is actually
wrong, file an issue or open a dedicated docs-only PR; do not edit it here.

# while this PR is open — GitHub drops the merge commit once it closes
git fetch origin e24f7f55664d7804b431f94b47735ebfccc65017 && git checkout e24f7f55664d7804b431f94b47735ebfccc65017
# afterwards, rebuild it from the two parents, which stay fetchable
git fetch origin f6b7c53db7b65bbfb019750efb4e545470b0c2b7 c5c2182f7643e1cb03969d546148ccc0aaeeee3e && git checkout -B drift-repro f6b7c53db7b65bbfb019750efb4e545470b0c2b7 && git merge --no-ff c5c2182f7643e1cb03969d546148ccc0aaeeee3e

node scripts/docs-audit/affected-docs.mjs --json f6b7c53db7b65bbfb019750efb4e545470b0c2b7

Advisory only, and a precision-first one (#9192): a page is listed because it names a
symbol, wire route or SDK method this diff touched — not because it mentions a changed
package. Each row says which anchor put it there, so a wrong row is reportable rather than
merely annoying. To re-verify, run the docs-accuracy-audit workflow scoped to these files:
node scripts/docs-audit/affected-docs.mjs f6b7c53db7b65bbfb019750efb4e545470b0c2b7 → pass the list as
args.docs, on the commit named under Which tree this was computed on.

Contract review (claude-fable-5-1, isolated seat) — PR #16887 @ c0d4ed6f9

Verdict: CHANGES REQUIRED. The code does what the ruling ordered — one table in packages/spec, all three producers resolve through it, no residual literal, no ALTER, no migration — and the SQLite pins are honest about what SQLite can measure. Two things the PR ships disagree with the repo: a protocol doc that now states the opposite of the new DDL, and a changelog sentence about MySQL that was never executed and that MySQL's documented conversion contradicts, while the live PG+MySQL harness the repo already runs in CI went unused.

Governed-surface check: docs/adr/**, .claude/**, skills/**, AGENTS.md, CLAUDE.md, content/docs/releases/** — none touched (git diff --name-only origin/main...c0d4ed6f9, 11 files, all under packages/** and .changeset/). packages/spec/src/conversions/registry.ts (#16693) not in the diff; the ADR-0113 sentence at the tail of createColumn is quoted, not edited. git merge-tree --write-tree origin/main refs/review/16887 → clean (origin/main = 142c01c88).

Findings

1. blocking — the protocol doc now states the opposite of the DDL, and was not edited. content/docs/protocol/objectql/types.mdx:266-268 — "SQL driver: a floating-point column (REAL on PostgreSQL/SQLite, FLOAT on MySQL). precision/scale are validation and display metadata — the DDL does not emit NUMERIC(precision, scale)"; :312 the same for currency; the mapping table at :1177 (number/currency/percent → REAL/FLOAT/REAL) and :1185 (summary → REAL). content/docs/references/api/sortability.mdx:68 — "summary is an engine-maintained table.float". After this PR a new column is numeric(65,30) on PG/MySQL and rating is integer. The Docs Drift bot on this PR says in its own caveat that types.mdx is exactly the page an emitter-only diff cannot list ("it was the page that diff falsified, in four places" — fix(driver-sql): emit varchar(maxLength) for a text field a declared index keys on #11430), so the bot's green is not coverage here. The PR's whole argument is "one declaration, one column"; the protocol doc is now a fourth answer. content/docs/protocol/** is hand-written and ungoverned (AGENTS.md → Documentation Guardrails row for content/docs/<tree>/), so it belongs in this PR, not in a rider. Same fossil, lower stakes: packages/services/service-analytics/src/measure-result-type.ts:161 still records "driver-sql's DDL answers col = table.float(name)" as the reason summary needs no correction.

2. blocking — the changeset asserts a MySQL refusal that was never executed, and MySQL's documented conversion says the opposite. .changeset/numeric-column-representation.md bullet 1: "rating is an INTEGER column on PostgreSQL and MySQL, so a fractional star count is now REFUSED there". The PR body's own gap 2: "MySQL was not executed … no MySQL server was driven." PostgreSQL refuses '4.5' into integer (text-param input syntax error). MySQL does not: a fractional value assigned to an integer column is rounded (4.5 → 5), with no error in strict mode, and a DECIMAL scale overflow is rounded with a Note (1265). That is a silent alteration on the dialect the changeset names as refusing — the exact class (A number field's declared scale is never enforced — values with more decimals are accepted and stored verbatim (min/max on the same field are enforced) #7501, "silent rounding is silently altering data") the table was chosen to end. The instrument exists in this repo: packages/drivers/driver-sql/src/live-dialect-matrix.testkit.ts (+ .globalsetup.ts) keyed on OS_TEST_POSTGRES_URL, used by 16 test files (e.g. sql-driver-aggregation-conformance.test.ts, sql-driver-column-order-dialects.test.ts), and run by the required check "Temporal Conformance (live PG + MySQL)" (.github/workflows/ci.yml:894, PG service + MySQL service conformance). Ask: either measure rating/DECIMAL on MySQL through that harness and pin it, or rewrite the bullet to say PostgreSQL refuses and MySQL rounds. A published changelog cannot carry a dialect claim the author marks unmeasured two sections later.

3. record — no live pin for the property the PR is about. All three new tests are better-sqlite3 (sql-driver-16318-numeric-representation.test.ts, generate-numeric-column-representation.pin.test.ts, spec pin) and say so honestly ("⛔ Do not read the oracle as a precision check"). The numeric(65,30) DDL, the information_schema three-producer table, the nullability four-shape table and the read-path string→number repair on PG are all PR-body prose from a session, reproducible by nobody from the tree. The env-gated harness in finding 2 is where a numeric_precision/numeric_scale read and a typeof === 'number' read belong; on SQLite the read-path pin at sql-driver-16318-numeric-representation.test.ts:139-149 cannot fire (SQLite never hands back a string for a float column), so the every-dialect move in formatOutput has zero automated coverage on the two dialects it was made for.

4. record — the read seam is binary64, and that bounds "money fidelity" end-to-end; the every-dialect pass also touches EXISTING columns. sql-driver.ts:17092-17099 does Number(v) on every string in numericFields on every dialect. Measured (node): for a value the driver itself wrote from a JS double the round-trip is exact — '1234567.890000000000000000000000000000' → 1234567.89, '12.500…' → 12.5 — because the shortest repr is what was stored. For anything not born as a JS double it is lossy and silent: '1234567890123456.123' → 1234567890123456; '9007199254740993.000…' → 9007199254740992; a full 30-scale '0.123456789012345678901234567890' → 0.12345678901234568. So the column is exact and find() is not: SQL-side writers, summary sums computed in SQL, anything ≥ 2^53. Not new as a wire limit (z.number().finite()), but the PR's "0/9 altered" is a column reading, not a find() reading, and the changeset should say the fidelity gained is exact-column-through-double. Second half: numericFields is seeded from NUMERIC_SCALAR_TYPES (sql-driver.ts:262), which includes the driver aliases integer/int/float for external/introspected columns. node-postgres returns int8/bigint as a string; before this PR that string left formatOutput untouched on PG, after it is Number()-ed — an existing external bigint column under a numeric field type silently rounds above 2^53 on PG. That is a read-path change to existing deployments outside "new tables only", unmeasured and unstated; state it or scope the pass.

5. record — the residual bound understates its onset. PR/spec docblock: "magnitudes below 1e-30 round to zero". True, but PG rounds any digit past the 30th place silently, so a full-precision double loses trailing significant digits from about |x| < 1e-13 (1.2345678901234567e-15 stores as 0.000000000000001234567890123457). The bound to state is "≤ 30 fractional digits are kept", not "below 1e-30".

6. record — ADR-0087 disposition rides the detector miss the gate's own docblock warns about. scripts/check-adr-0087-registration.mjs:141-147: no-migration-prescription is refused only when the body carries a prescription, and :199-206 (fix(service-package): classify a publish driver fault as 5xx and stop returning driver text as caller data (#8131) #8277) says an exemption held because the detector found nothing is the failure mode. Changeset bullet 3 is a prescription in substance: "A generated migration no longer emits NOT NULL for a field marked only required: true. Declare storage: { notNull: true } for a physical constraint" — a FROM (required: true) → TO (storage.notNull) edit a consumer must make to keep the column they had, written without the arrow/table framing the detector matches. "Nothing authorable moves" is true of keys; the output of os generate migration for an unchanged input moves (types and NOT NULL), which is what a consumer with checked-in migrations diffs. Either argue the exemption positively in the marker (the ledger is for migrate meta rewrites; this is an addition, not a rename — defensible) or register. Check Changeset is green on all matrix legs, consistent with a miss rather than a finding.

7. observation — ruling compliance, itemised. (a) One table in packages/spec ✓ numeric-column-representation.ts:196-233. (b) Three producers read it ✓ — sql-driver.ts:16483-16489, generate.ts:1134-1136,1212,1243-1245 via numericSqlType, generate.ts:1985-1992; grep of both files at the ref finds no DECIMAL(/decimal(/float( literal for a FieldType: the remaining table.float (:16436) and table.integer (:16429) arms are the driver aliases float/integer/int, which are not in the FieldType enum (field.zod.ts:40-100). (c) New tables only ✓ — the three alterTable sites (:10128,12193,12462) add columns through createColumn; no type alter, no drift reader (schema-drift.ts has no numeric expectation; nullability drift reads storage.notNull, :849). (d) Both pre-work measurements are in the PR body, with controls; the 5,2 inference corrected to rounding. (e) rating integer with the half-star search stated in changeset ✓. (f) Nullability landed here rather than paired — the ruling says "same PR or a paired one", the dispatch says "the dev's call, stated either way"; stated ✓, within the ruling. (g) Deviation worth naming: the ruling asked per-type answers ("precision chosen for money", "a stated scale", "a scale that does not truncate"); the table is per-type in shape and uniform in value — 65,30 for all six, justified as the portable maximum. It satisfies each constraint; it is not what "chosen for money" reads as, and the maintainer should see that it was answered with "the ceiling".

8. observation — totality and the pin. The pin (numeric-column-representation.test.ts:25-27) enumerates from NUMERIC_VALUE_TYPES, not a second hand list ✓, and NUMERIC_VALUE_TYPES is satisfies readonly FieldType[] (field-value.zod.ts:63-65) — a subset check, itself a hand list. A numeric type added to the FieldType enum but not to that set still falls to default → table.string in the driver; pre-existing, out of this table's reach, worth a sentence. summary belongs in the table (it is stored, was table.float, is in NUMERIC_VALUE_TYPES); "no seam" is true as stated — COMPUTED_VALUE_TYPES excludes it from validateRecord's type door — and the aggregateSummaryValue verbatim min/max over a text child (measure-result-type.ts) is now refused by an exact column exactly as real refused it.

9. observation — numericSqlType runs at module evaluation. FIELD_TYPE_SQL_MAP is a module const (generate.ts:1134), so a missing entry throws when generate.ts is imported — every os command, not only generate migration. Loud by design; the blast radius is the CLI, and the spec pin fails first in CI.

10. observation — tests, surface, hygiene. No .skip/.only/.todo in the diff. Ablation: NUMERIC_COLUMN_SCALE 30→2 fails exactly toBe(30) / toBeGreaterThan(2) / >= 5 in one it (:57-64); the other four compare against the constants and stay green — "1 failed / 4 passed" follows. generate-multiple-json-column.pin.test.ts keeps its subject (multiple does not decide nullability) and adds the required-only-is-nullable half. api-surface/data.json / export-origins/data.json: additions only — NUMERIC_COLUMN_PRECISION, NUMERIC_COLUMN_REPRESENTATION, NUMERIC_COLUMN_SCALE (const), NumericColumnRepresentation (type), numericColumnFor (function). Changesets minor ×3 with BREAKING banner; check-changeset-no-major has nothing to refuse; the packages/drivers/driver-sql/src/ level-axis blind spot ([finding] The changeset LEVEL axis is blind to every NESTED package: packages/*/src/** matches one segment, so 51 of 74 workspace packages (all drivers/services/adapters) can pair Clause-②: yes with patch and stay green #16713) is declared and the grade is the author's. Fixes #16318 in the body only; seven commits, none carries a closing trailer. CI at c0d4ed6f9: 55 of 56 check runs completed, all success or skipped (Build Core, Test Core 1–6, Temporal Conformance, Dogfood ×4, Type Check ×4, Check Changeset ×6, Governed Surface Queue Guard, Spec property liveness); "Lint & Repo Gates" still in_progress at read time. mergeable_state: blocked — draft, needs:contract-review with no approving review, one required check unfinished.

Maintainer-only merge: yes. This decides what every NEW table gets for money, percent and rating on PostgreSQL and MySQL — numeric(65,30) read through a JS double, integer for stars — under a ruling recorded as the director seat's reading of 「继续决策」 with "the maintainer may overturn on a word"; it is cross-lane (engine + cli), Clause-②: yes, and finding 2 puts a dialect claim in a shipped changelog that the maintainer's own MySQL deployments would contradict. The maintainer's eyes are warranted, and finding 1 and 2 should be fixed before they look.

Generated by Claude Code

Director seat adoption record — summon #20, session_01Tep4AYXZvyBA7jsvne5KZV (os-bill), 2026-09-09T06:59Z. The verdict below is adopted verbatim from an isolated contract-review subagent (explicit model = CONTRACT_REVIEW_TIER). Transcript tier check before adoption: every harness-stamped model field in the subagent transcript reads claude-fable-5-1 (95 stamps, no other value). Head re-read at posting time = 034856799a, unchanged since the review. ⛔ This seat takes no release action on this carrier (no ready flip, no auto-merge, no enqueue, no label write): the owning seat (domain:engine / consolidated seat (takeover claim 5594621006)) adopts this verdict verbatim or discards it, and acts per the state machine.

Contract review (CONTRACT_REVIEW_TIER, isolated seat) — PR #16887 @ 034856799a4107a5c965be47f99ae2cf4d023877

Verdict: PASS WITH FINDINGS

Both blocking items from the prior review (5586825380 @ c0d4ed6f9, carried on the card as 5586839197) are discharged on this head, and four of the six items are discharged in full; F5 is discharged in the changeset but not in the spec docblock the changeset points readers at. Nothing found on the head moves a published contract beyond what the prior review approved. Ruling C ∩ ④ (batch #86, 5581958053) is still executed in substance: one table in packages/spec, three producers resolve through it with no residual literal, new tables only, nullability from storage.notNull in both formats. Head = c0d4ed6f9 + merge of origin/main @ ee2cb6b40 (1be05800c7) + one patch commit (034856799a, session session_018rzQyhLGC5iVs11V3TzRs5); increment = 16 files (git diff origin/main...refs/pr-review/16887); delta since the reviewed head = 7 files (git diff 1be05800c7..034856799a).

Owed items from the prior review

1. F1 (blocking) — docs falsified, unedited → DISCHARGED. content/docs/protocol/objectql/types.mdx:266-281 (number: exact decimal NUMERIC(65,30)/DECIMAL(65,30)/float, "New columns only", the field's own precision/scale still not emitted), :322-335 (currency, with the 0-/3-digit-currency reason for not fixing two decimals), :353-357 (percent), matrix rows :1200-1201 (family + new rating row) and :1209 (summary), footnote :1236-1251 (new-columns-only, SQLite affinity, PG refuses / MySQL rounds). content/docs/references/api/sortability.mdx is AUTO-GENERATED — DO NOT EDIT (its :5), so the head fixed the source packages/spec/src/api/sortability.zod.ts:64-71 and regenerated :68-75 — I compared the "Considered and deliberately NOT members" paragraph of the docblock (comment prefix stripped) with the mdx: byte-identical, so the next gen:docs will not revert it. Fossil packages/services/service-analytics/src/measure-result-type.ts:159-173 corrected (comment-only). No content/docs page on the head still says REAL/FLOAT/table.float for the family except the corrected sortability.mdx:68.
2. F2 (blocking) — unexecuted MySQL refusal claim → DISCHARGED. .changeset/numeric-column-representation.md:41-46 now states PostgreSQL REFUSES 4.5, MySQL ROUNDS to 5 "with no error", SQLite unchanged. Pinned per dialect in packages/drivers/driver-sql/src/sql-driver-16318-numeric-representation-live-dialects.test.ts:189-208 (PG: 'refused' and 0 rows; MySQL: 'accepted' and f_rating === 5, not.toBe(4.5)). Execution evidence: see F5 below (step-level in CI; the sibling line's independent live MySQL 8.0.46 reading in 5595029333 agrees).
3. F3 (record) — no live pin → DISCHARGED. New 281-line cell on the existing harness: catalog numeric_precision/numeric_scale compared to numericColumnFor rather than a literal (:124-154, with a non-vacuity check that the catalog answered and that two distinct arms exist), typeof === 'number' read seam (:156-175), the 1234567.89 / 33.333 discriminators (:177-187), rating per dialect (:189-208), and the F4 scope cell (:235-281). It runs under the "Temporal Conformance (live PG + MySQL)" job: .github/workflows/ci.yml:1170-1183 sets both URLs and OS_EXPECT_LIVE_DIALECT_MATRIX: '1' and runs pnpm --filter @objectstack/driver-sql test = vitest run with no custom include/exclude (vitest.config.ts carries only globalSetup); declareDialectCell (live-dialect-matrix.testkit.ts:506-511) routes an unavailable cell (available: !!MYSQL_URL, :313) to declareUnprovisionedCell → expect.fail under the flag (:372-380), so a missing server is red, not a skip.
4. F4 (record) — read seam binary64 / pass reaches existing columns → DISCHARGED by scoping, not only by statement. sql-driver.ts:4390-4413 new protected numericValueFields (the NUMERIC_VALUE_TYPES half without the integer/int/float aliases); populated on every registration path — shard alias :9669, registerExternalObject :9810/:9823, initObjects registration :9895/:9920 (all three numericFields write sites have a mirror; no orphan); selected per dialect at :17168-17170 (SQLite keeps the wide set for the legacy TEXT-affinity repair). The binary64 bound is stated at :17151-17164 and in the changeset :53-59. Live PG pin :244-280: an external bigint holding 2^53+1 under an alias-typed field reads back as the unrounded string while the numeric(65,30) column in the same row is coerced to 12.5.
5. F5 (record) — residual bound understated → PARTIALLY DISCHARGED. Changeset :47-52 now says "keeps 30 fractional digits … loss begins around |x| < 1e-13 and is total below 1e-30". The spec docblock packages/spec/src/data/numeric-column-representation.ts:100-106 still says only "Magnitudes below 1e-30 round to zero" — see Finding F1.
6. F6 (record) — ADR-0087 disposition on a detector miss → DISCHARGED. Marker at changeset :70 argues positively; its load-bearing claim checks out on origin/main: packages/spec/src/conversions/registry.ts:2021-2050 carries the field-required-notnull-explicit withdrawal tombstone (ADR-0087's field-required-notnull-explicit conversion asserts an implication ADR-0113 abolished — the boot calls it a forward conversion, but its output and the source it prescribes disagree at the storage layer #16693, batch Add comprehensive protocol review and Q1-Q4 2026 optimization roadmap #85, "No migration is owed to anyone"). node scripts/check-adr-0087-registration.mjs --base origin/main --head refs/pr-review/16887 → exit 0. The marker's internal pointer is wrong — see Finding F2.

Derived judgments

• packages/spec/src/data/numeric-column-representation.ts (new): five new exports on @objectstack/spec/data — NUMERIC_COLUMN_PRECISION, NUMERIC_COLUMN_SCALE, NUMERIC_COLUMN_REPRESENTATION (consts), NumericColumnRepresentation (type), numericColumnFor (function); api-surface/data.json and export-origins/data.json are additions only (+5 each). index.ts:222-227 one export *. Public surface widens → Clause-②: yes is right. No new key on any payload, no new error code, no enum member (FieldType untouched; the table is Record<string, …> with membership pinned by numeric-column-representation.test.ts:25-27 in both directions against NUMERIC_VALUE_TYPES).
• packages/spec/src/api/sortability.zod.ts:64-71: docblock only; no schema moves.
• packages/drivers/driver-sql/src/sql-driver.ts: behaviour of a published door moves — createColumn numeric arm :16514-16533 resolves the seven NUMERIC_VALUE_TYPES through numericColumnFor (table.integer / table.decimal(name, 65, 30)), the old summary table.float arm is removed (old :16522-16527), the float alias keeps table.float (:16476-16478); formatOutput numeric coercion now runs on every dialect (:17123-17178), which also repairs EXISTING external objects mapped onto numeric/DECIMAL columns (stated in the PR body). New protected member numericValueFields is subclass-visible surface. New tables only: the three alterTable sites still add columns through createColumn; no type alter; schema-drift.ts unchanged in this diff.
• packages/cli/src/commands/generate.ts: published CLI output moves — FIELD_TYPE_SQL_MAP entries :1134-1136, :1212, :1243-1245 via numericSqlType (:1307-1316, throws at module evaluation on a missing entry: every os command that imports generate.ts, loud by design, the spec pin fails first); declaredNotNull :1330-1332; NOT NULL from storage.notNull in --format sql :1771 and typescript :1888; typescript numeric arm :1979-1994. generate-numeric-column-representation.pin.test.ts derives every expectation from numericColumnFor (:108, :133, :149, :166-167) and asserts the old literals are gone (:138-139).
• packages/services/service-analytics/src/measure-result-type.ts:159-173: comment-only; no behaviour, no changeset owed.
• packages/lint/**, packages/metadata-protocol/**, packages/objectql/**: not in the diff (see Finding F3).
• Beyond-prior-approval check on the delta: numericValueFields narrows the pass the prior review asked to be scoped; everything else in 034856799a is docs, comments, a test file and changeset prose. Nothing moves further than approved.

Semver / changeset

.changeset/numeric-column-representation.md: @objectstack/spec, @objectstack/driver-sql, @objectstack/cli all minor — a widening takes at least minor, major refused in the launch window. node scripts/check-changeset-no-major.mjs --base origin/main --head refs/pr-review/16887 → "introduces no major bump", exit 0 (level axis reported N/A offline for lack of a PR payload; the CI "Check Changeset" run on the head is green and the axis needs at least one moved package graded minor, which three are). BREAKING banner present at :38. ADR-0087 marker present at :70, gate exit 0. The sibling line's "3 → 7 packages" is not owed by any rule: comment-only edits in lint/metadata-protocol/objectql/service-analytics release nothing, and none is graded patch (the level-axis prohibition). Governed paths in the 16-file increment: none (.claude/**, skills/**, docs/adr/**, AGENTS.md, CLAUDE.md, content/docs/releases/** all untouched; the skills/** rows in a two-dot diff are main's, arriving via 1be05800c7).

Boundary flags

• Cross-lane by design: packages/cli/** (domain:cli) under a domain:engine claim — declared in the dispatch (5584575074) and the PR body.
• content/docs/protocol/** is hand-written and ungoverned; in-PR is correct. content/docs/references/api/sortability.mdx is generated; edited through its source — the prior review's instruction to edit the .mdx line directly would have been reverted by the next gen:docs, and the head did the right thing.
• ADR-0087's field-required-notnull-explicit conversion asserts an implication ADR-0113 abolished — the boot calls it a forward conversion, but its output and the source it prescribes disagree at the storage layer #16693's packages/spec/src/conversions/registry.ts is not in the diff; after the merge the ADR-0113 block in sql-driver.ts is main's fix(spec): withdraw the ADR-0087 field-required-notnull-explicit conversion — required: true stops stamping storage.notNull #16890 text and this PR's comment additions sit in the numeric arm, not in that block. git merge-tree --write-tree origin/main refs/pr-review/16887 (main @ 854639b3, feat(engine)!: findOne, update and delete declare what they answer, and their hook seams are guarded (#16231) #16783 landed) → clean; API mergeable: true / clean.
• Ownership: card #16318 is pm:queue, assignee empty, needs:contract-review on both carriers; the last Claim: on the card is the consolidated seat's takeover (5594621006, 02:00:11Z), whose dev was killed by a 429 at ~03:29Z (5597041636) after pushing 034856799a with no os-dev-report. The sibling branch claude/issue-16318-patch-dup-01ADLdAs @ 5339717b6d was NOT reviewed here; its report was read as claims only.
• Dev report (5595029333, the latest on the card) — open questions answered. Q1 (which line lands): A — keep 034856799a; it is the head under review, CI-green, and F1–F6 are discharged on it as verified above. Take from 5339717b6d only the five comment fossils (Finding F3); do NOT take the four extra changeset packages (not owed); the F2 wording upgrade is already on the head (:43-44 "with no error"). Q2 (double dispatch or resumed session): a deliberate takeover claimed on the card — 5594621006 is an explicit Claim: posted before the domain:engine seat's ~02:1xZ re-dispatch, and the head commit's trailer names that claim's session; per CLAUDE.md the newest Claim: decides, so the stand-down in 5595053286 was the correct outcome. deviations: the report declares none (no key); its out_of_scope_findings on the generated-mdx hazard is confirmed on the head; its "both servers available in this container" is an environment claim I did not need and did not verify.

Findings

• F1 — non-blocking. packages/spec/src/data/numeric-column-representation.ts:100-106 still states the residual bound the prior review's F5 corrected ("Magnitudes below 1e-30 round to zero"), while the changeset :47-52 states "30 fractional digits are kept … loss begins around |x| < 1e-13". The docblock says of itself "stated here so no reader has to rediscover it", and sql-driver.ts:16487 and generate.ts both tell readers the spec module "carries the residual bound" — so the canonical home now disagrees with the shipped changelog. One sentence.
• F2 — non-blocking. Changeset self-references: :39 "Three consequences" above four bullets (:41, :47, :53, :60); the ADR-0087 marker :70 says "bullet 3 IS a prescription", but bullet 3 (:53) is the read-seam bullet and the prescription is bullet 4 (:60). The marker is the auditable artifact the gate prints verbatim; a wrong pointer defeats its purpose. Fix with F1 in one commit.
• F3 — non-blocking. The same summary-is-table.float sentence corrected at sortability.zod.ts:64 still stands in five comment sites the PR does not touch: packages/metadata-protocol/src/protocol.ts:3087 (docblock table) and :9400; packages/lint/src/validate-sortable-fields.ts:70; packages/lint/src/validate-sortable-fields.test.ts:27, :105; packages/objectql/src/query-expression-conformance.test.ts:92, :602; packages/services/service-analytics/src/__tests__/measure-result-type.test.ts:189, :666. Code comments, no published page (verified by grep of content/docs on the head) — the [finding] Seven out-of-package comments still say formatOutput's audit and datetime repairs sit inside if (this.isSqlite) — false since PR #16619 landed, and one of them tells readers a declared Field.datetime is NOT protected on Postgres/MySQL #16728/[finding] Eleven more out-of-package comments still gate formatOutput's timestamp repairs on if (this.isSqlite) — the same drift as #16728, invisible to its normalizeSqliteDatetimeOutput census #16818 class, not the F1 class. Sweep here (the sibling 5339717b6d carries it) or file a follow-up; either is acceptable.
• F4 — non-blocking. The PR body is behind the head: it pins every reading to c0d4ed6f9f, lists an 11-file surface (the head has 16) and omits the live-dialect cell, sortability.zod.ts/.mdx, types.mdx and measure-result-type.ts; no os-dev-report exists for the head's patch round. The body is what the maintainer reads at merge time; it must be brought to the head by whoever re-claims.
• F5 — record. Evidence that the live cells executed on MySQL is step-level, not log-line-level: the Temporal job on the head (102317539991) completed success with step 12 "Run driver-sql suite against both live servers" success, under the flag and URLs cited above, and the harness makes a missing server red. The per-file log line was not retrievable from this seat (the REST log redirect targets productionresultssa*.blob.core.windows.net, which the proxy answers 403 CONNECT; the MCP log tail is capped at ~600 KB and begins inside step 14). The head's own commit message says the MySQL half was run only in CI; the sibling line's local MySQL 8.0.46 reading (4.5 → 5, 4.4 → 4, SHOW WARNINGS empty) agrees with the pinned expectation.
• F6 — observation. No .skip/.only/.todo in the diff; no throw introduces a registered error code; sql-driver-16318-numeric-representation.test.ts (SQLite) and the spec pin are unchanged since c0d4ed6f9 and still derive from spec constants. Per-type SQLite consequences are stated in the spec module, the driver arm (sql-driver.ts:16494-16512) and types.mdx:1245-1251.

Maintainer-only merge: yes — unchanged from the prior review's reading (feat!, Clause-②: yes, cross-lane, decides what every new table gets for money/percent/rating on PG and MySQL under a 「继续决策」 ruling). Per 5586839197, on this PASS the card goes pm:awaiting-maintainer and the PR stays draft; F1/F2 are a one-commit tidy the maintainer may want before reading, and do not gate.

CI at read time

Head 034856799a4107a5c965be47f99ae2cf4d023877: 38 check runs, 34 latest-per-name — 30 success, 4 skipped (Auto Label, Check PR Size, Console Pin Gate, Packed-tarball smoke (opt-in)), 0 failed, 0 in progress; every run's head_sha is the head. Green includes Build Core, Test Core (+6 shards), Temporal Conformance (live PG + MySQL), Type Check ×5, Check Changeset, Lint & Repo Gates, Governed Surface Queue Guard, Spec property liveness, Dogfood ×5, both single-writer/same-issue guards. PR: draft, mergeable_state: clean, labels documentation, size/xl, tests, tooling, protocol:data, needs:contract-review.

Implemented-by: branch claude/issue-16318-numeric-column-representation-table
Reviewed-by: director seat summon #20 (isolated fable subagent, transcript-verified before adoption)

{"pr":16887,"head":"034856799a4107a5c965be47f99ae2cf4d023877","verdict":"PASS WITH FINDINGS","blocking":[],"clause2":"yes","semver_ok":true,"governed":false,"ci":"38 runs / 34 latest-per-name: 30 success, 4 skipped (Auto Label, Check PR Size, Console Pin Gate, Packed-tarball smoke opt-in), 0 failed, 0 pending; mergeable clean"}

Generated by Claude Code

Landing note — the standing PASS WITH FINDINGS (5597570687, head 034856799a) carries to head c5c2182f7643e1cb03969d546148ccc0aaeeee3e; carriers come off both faces; maintainer-only merge — the PR stays draft and #16318 goes to pm:awaiting-maintainer (director seat, summon #18 segment 3, session_017Js5kTpTtxieBjPyScgxJ3)

Delta verified by this seat at CONTRACT_REVIEW_TIER (git diff 034856799a..c5c2182f): 2 files, +15/−8 — packages/spec/src/data/numeric-column-representation.ts docblock only (git diff -U0 | grep '^[+-]' filtered for non-comment lines → none; the residual-bound section now states the changeset's own bound: 30 fractional digits kept, precision loss begins around |x| < 1e-13, total below 1e-30, magnitudes ≥ 1e35 refused) and .changeset/numeric-column-representation.md prose only (:39 「Three」 → 「Four consequences」; the ADR-0087 marker now points at bullet 4, the NOT NULL one, and remains the single marker — count 1 on the head). No export, schema, generated artifact or test moved; the spec-file hunk is docblock prose, so the accept-set findings of the original verdict are unchanged and no re-review is owed. ⇒ Verdict F1, F2 discharged; F4 discharged by the PR body edit (re-pinned to c5c2182f, the 16-file surface listed incl. the live-dialect cell, sortability.zod.ts/.mdx, types.mdx, measure-result-type.ts; the body's own understated residual-bound paragraph and its stale 「MySQL was not executed」 gap corrected in the same edit) and the head's os-dev-report at #16318 5599295515 (adr-0087 gate 0 with the corrected marker printed, changeset-no-major 0, empty-changeset 0, spec numeric pin 5/5, spec typecheck 0, check:docs 0 「228 generated files in sync」, check:authorable-surface 0, eslint 0/0).

Not in this round, by scope: F3 — the same falsified 「summary is a float column」 sentence in five comment sites outside this PR's files (metadata-protocol/src/protocol.ts:3087, :9400; lint/src/validate-sortable-fields.ts:70 + test :27, :105; objectql/src/query-expression-conformance.test.ts:92, :602; service-analytics/.../measure-result-type.test.ts:189, :666). The remote still carries a sibling branch claude/issue-16318-patch-dup-01ADLdAs @ 5339717b6d (no PR; the domain:engine seat's earlier attempt) that the round reports already holds that sweep — the maintainer or the engine seat may cherry-pick it after this merges; it is not owed here. F5/F6 unchanged.

Recorded deviation, accepted: the commit trailer names the model that did the round (a trailer is provenance; AGENTS.md's attribution clause permits exactly that form).

Chain: needs:contract-review off this PR and off #16318; #16318 pm:dispatched → pm:awaiting-maintainer (the one remaining act is the maintainer's merge: protocol:data); PR stays draft — ⛔ not readied, not enqueued. CI on c5c2182f at this reading: 17 success / 16 in progress / 4 skipped / 0 failure; the branch is ~45 commits behind origin/main but git merge-tree is clean and the queue rebuilds on merge.

Implemented-by: claude/issue-16318-numeric-column-representation-table (tail round under claim 5598825650) · Reviewed-by: 5597570687 (isolated claude-fable-5-1, summon #20) + this seat's delta reading.

Generated by Claude Code