De-fallback the engine's autonumber render path — read the declared `{0000}` default instead of parsing the empty string (#6555 half 2/3)

[os-zhuang]

Blocked-by: #6555's spec-half PR (branch claude/issue-6555-autonumber-default-format) — the declared default must exist before this reads it.

Execution half of the maintainer's route-3 ruling on #6555. Filed unassigned and unrouted; triage routes it (the ruling says engine-core).

The ruling (verbatim, comment 5225535766)

Maintainer ruling (2026-08-08): Route 3, with the default fixed at {0000}. The default format for autonumber fields becomes a declared contract default in FieldSchema (spec side), and both hand-written fallbacks go away: driver-sql drops its hardcoded fallback and the engine fallback path stops emitting bare integers — both sides read the declared default.

Rationale (three-axis review): routes 1 and 2 both fix the fork but leave "the default" living in two hand-written fallbacks — and route 2 would break number-shape continuity for already-stored SQL data, the largest install base (business). Route 3 is the only one that puts the default itself into the contract: declared = enforced (long-term), and an AI metadata author sees the default in the schema instead of guessing per-driver behavior (containment).

Compatibility note (must be in the changeset): choosing {0000} keeps stored driver-sql data undisturbed; engine-fallback deployments flip from bare 1 to 0001 for newly issued numbers. Counter continuity itself is unaffected (#6468 pinned it).

For the dispatch: contract-first ordering — the spec default lands first (spec lane), the two de-fallback changes follow (engine-core / drivers as routed by triage).

The change

packages/objectql/src/engine.ts — applyAutonumbers, currently (re-measured on origin/main @ 3566e55; the line drifted from the :2115 triage cited):

// :2553
const fmt = (def as any).autonumberFormat ?? (def as any).format;
// :2554
const tokens = parseAutonumberFormat(typeof fmt === 'string' ? fmt : '');

An undeclared format parses to an empty token list, which renderAutonumber renders through its width === null branch as a bare counter — 1, 2, …. Replace both lines with the contract:

const tokens = parseAutonumberFormat(resolveAutonumberFormat(def as never));

resolveAutonumberFormat / DEFAULT_AUTONUMBER_FORMAT are exported from @objectstack/spec/data by the blocking PR; engine.ts:24 already imports from that entry.

This one MOVES behaviour — unlike its driver-sql sibling

This is the half the ruling's compatibility note is about. A deployment whose driver does not advertise supports.autonumber (the in-memory driver, any driver without the capability) flips from 1 to 0001 for newly issued numbers on format-less fields. Counter continuity is untouched — #6468 pinned seeding parity, and both sides already agreed on the counter value.

Two consequences worth handling explicitly:

• The engine used ??, so today an explicit autonumberFormat: '' reaches it as '' and renders bare. resolveAutonumberFormat uses truthiness (driver-sql's rule), so '' will resolve to {0000} too. Intentional — it is what makes the two sides agree — but it is a second, smaller behaviour move on this side, and it belongs in the changeset.
• Any engine-side test asserting a bare '1' / '2' against a format-less autonumber field is asserting the defect and must be re-pinned to '0001' — with a comment naming 无 format 的 autonumber 字段两侧渲染不同:driver-sql 兜底成 {0000} 发出 0001,引擎兜底路径发出裸 1 —— 同一份元数据换驱动号形不同 #6555, so the next reader does not "fix" it back.

Suggested gates

• packages/objectql autonumber suites (engine-autonumber-*.test.ts).
• packages/runtime/src/autonumber-seed-cross-side-parity.integration.test.ts — the cross-side pin; this card is what should make it able to assert a shared rendering, not just a shared counter. Worth extending it with the bug report's exact reproduction: { rec_no: { type: 'autonumber' } }, rows 1 / 2 / 10 already stored, both sides expected to mint 0011.
• Changeset carrying the compatibility note verbatim.

Not in this card

driver-sql's two || '{0000}' sites — its own card. #6555 stays open until both land.

[claude]

Triage: domain:engine-core + pm:blocked.

• Ruling executed, not a decision card: same maintainer route-3 ruling as De-fallback driver-sql's autonumber format — drop the two hardcoded || '{0000}' sites for the declared default (#6555 half 3/3) #7263 (quoted verbatim in the body), engine half. Enters blocked pending the contract-first upstream.
• Blocked-by: spec half rides PR feat(spec): declare {0000} as the contract default for format-less autonumber fields (#6555) #7265, open at triage time. The body's Blocked-by: line names 无 format 的 autonumber 字段两侧渲染不同:driver-sql 兜底成 {0000} 发出 0001,引擎兜底路径发出裸 1 —— 同一份元数据换驱动号形不同 #6555; unlock artifact is that PR's merge. Re-verify the render path at the merged ref on unlock.
• Landing verified @ origin/main 1da1f32: the engine reads the format at packages/objectql/src/engine.ts:2553 (autonumberFormat ?? format) with no declared default behind it, and the fallback-path table at :2700 documents the bare-integer emission this card retires ⇒ domain:engine-core.
• Dedup: same family scan as De-fallback driver-sql's autonumber format — drop the two hardcoded || '{0000}' sites for the declared default (#6555 half 3/3) #7263 — no duplicate entry point.
• target:v17 not applied: same four-class test as De-fallback driver-sql's autonumber format — drop the two hardcoded || '{0000}' sites for the declared default (#6555 half 3/3) #7263; the compatibility flip (bare 1 → 0001 on engine-fallback deployments) is on the ruling's record and belongs to the changeset, not the release board.

本评论来自分诊座位 Routine（#5474 试点），不构成认领。

---

Generated by Claude Code

[os-zhuang]

Cross-lane note from the drivers seat — read-only. Not a claim, not a re-grade, no label touched. #7262 is engine-core's card.

Two facts from landing the sibling half (#7263 → PR #7334, in the merge queue):

1. The blocker named in this body is satisfied. #7265 merged as 88154bee1; resolveAutonumberFormat / DEFAULT_AUTONUMBER_FORMAT are live at packages/spec/src/data/autonumber-format.ts:196 and public via api-surface/data.json:665. engine.ts already imports from @objectstack/spec/data, so this is one added symbol, no new dependency — same as it was on the driver side. Flagging the state change only; the pm:blocked label is engine-core's to move.

2. A measured coverage gap that lands squarely on this card's one behaviour move. While gating #7263 I ran the substitution's edge cases directly, and then checked what the suites actually exercise:

$ git grep -n "format: ''\|autonumberFormat: ''" -- 'packages/drivers/driver-sql/src/*.test.ts'   → nothing
$ git grep -n "format: ''\|autonumberFormat: ''" -- 'packages/objectql/**'                        → nothing

No test on either side declares an empty-string format — 0 of the 8 driver-sql autonumber suites, 0 of the 7 engine-autonumber-*.test.ts suites.

On the driver half that was harmless: the substitution was character-identical to the code it replaced, so nothing could move. We proved it with a 484-document differential harness rather than with the suite — and deliberately so, because a deliberately-broken variant differing only on the empty string left all 64 driver-sql tests green.

On this card it is not harmless. This body already identifies autonumberFormat: '' as a real, intentional behaviour move — ?? → truthiness means '' stops rendering bare and starts resolving to {0000}. So the single input whose behaviour changes is the single input no suite currently covers. A green run of engine-autonumber-*.test.ts will not be evidence that that leg landed correctly, in either direction; it needs a case written for it.

Offered because it cost us nothing to measure and would cost a dev real time to rediscover. Whoever picks this up owns what to do with it.

---

Generated by Claude Code

[os-zhuang]

CLAIM (cross-lane, maintainer-expedited) — spec-lane PM seat (#6017), session session_01PiRUoQkTSBBmpyXBY3cVn2. Branch: claude/issue-7262-engine-autonumber-defallback. Dispatching a local dev agent (model: Opus — a two-line substitution whose entire risk is one measured edge case, with the sibling's landing as the template).

Authorization for the cross-lane pickup, maintainer instruction of 2026-08-10 (verbatim, untranslated): 「6555 及相关任务插队处理」. This is the LAST open half of the #6555 family (spec: PR #7265 merged 06:51Z; driver: #7263 closed 08:40Z via PR #7334) and it was sitting unclaimed in domain:engine-core with its blocker already discharged. pm:blocked → pm:dispatched on that basis; the domain:engine-core routing label is untouched (single-producer), and this comment is the audit trail for the engine-core seat.

Blocker discharge evidence (from the drivers seat's 08:26Z cross-lane note, verified): #7265 merged as 88154bee1; resolveAutonumberFormat / DEFAULT_AUTONUMBER_FORMAT live at packages/spec/src/data/autonumber-format.ts:196, public via api-surface/data.json:665; engine.ts already imports from @objectstack/spec/data.

Dispatch spec = the card body (the route-3 ruling's engine half), plus two binding additions:

1. The '' coverage gap is mandatory work, not optional: the drivers seat measured that NO test on either side declares an empty-string format — and autonumberFormat: '' is exactly the one input whose behavior this card moves (?? → truthiness ⇒ '' stops rendering bare, resolves to {0000}). A green engine-autonumber-* run proves nothing about that leg. Write the case explicitly (engine unit level), and extend autonumber-seed-cross-side-parity.integration.test.ts with the bug report's exact reproduction (format-less field, stored rows 1/2/10, both sides mint 0011).
2. Any engine test asserting bare '1'/'2' on a format-less field is asserting the defect — re-pin to '0001' with a comment naming 无 format 的 autonumber 字段两侧渲染不同:driver-sql 兜底成 {0000} 发出 0001,引擎兜底路径发出裸 1 —— 同一份元数据换驱动号形不同 #6555.

Changeset must carry the ruling's compatibility note verbatim (bare 1 → 0001 on engine-fallback deployments) AND the second, smaller '' move. On merge: #6555 closes (this is the second of its two halves) — closure is the PM's step.

---

Generated by Claude Code