fix(objectql): 批量写上引用 previous 的 hook 条件 —— 点名「当前版本限制」的专门诊断 + ADR-0058 按行语义补遗 (#5037) by xuyushun441-sys · Pull Request #5057 · objectstack-ai/objectstack

xuyushun441-sys · 2026-08-04T01:13:40Z

维护者 2026-08-04 对 #4800 / #4862 的拍板里,本单是 rc 止血半步:契约(批量写按行求值、按行触发)记进 ADR,实现留给 #5038,rc 期把撞上限制时的报错从「像是作者写错了」改成「点名这是当前版本的限制」。

先说清楚现状:第 1 项已部分落在 #4861 里

派单文里描述的「批量写 + 引用 previous → 抛泛化的 No such key: previous」在 origin/main 上已不成立:PR #4861 在实现 #4775 的同时就带了 predicateBulkWrite 分支。实测(worktree 内跑现有用例)当前报错是:

Hook 'bulk_breaker' could not evaluate its condition (type: Unknown variable: previous)
— operation aborted. The condition reads 'previous', but this is a PREDICATE bulk write
(multi: true): ... Rewrite the condition without 'previous', or target the write at one
record (update by id). ...

所以本 PR 不是「补一条没有的诊断」,而是把这条诊断对齐到 8-04 的裁定,并补上裁定要求而当时没有的三样东西:契约口径、机器可读判别位、ADR 记录。这一点写在这里,免得 review 时按派单文原样核对。

改了什么

1. 文案:从「换个写法」改成「这是当前版本的限制」

旧文案第一条出路是 rewrite the condition without previous。在 8-04 的裁定之后这句话是错的建议:作者写的过渡条件是合法的、平台已声明要支持的形状;去掉 previous 会把「刚变成 done」悄悄变成「是 done」,于是本来只该记一次的审计,在每一行早就 done 的记录上都会触发 —— 一个静默的语义改变,正是 #4649/#4775 一路在清的那类东西。

现在的文案:点名批量写与「没有单一前置记录」(保留),然后说明这是当前版本的限制而非契约、引用契约出处(ADR-0058 补遗、#4800/#4862)与退役条件(#5038),先给今天可用的出路(改单记录写入,同一条件原样成立),再把「删掉 previous」标价而不是推荐。原有的「record-change flow trigger 不是绕路」一句保留 —— #4862 是实测结论,它订的是同一批生命周期 hook,批量上同样拿不到 previous。

2. 判别位:`limitation`,并且故意不叫 `code`

HookConditionError 新增 limitation?: 'bulk_write_previous_unbound' | 'bulk_write_stored_state_unavailable'(类型 HookConditionLimitation 一并导出),与既有的 reason / predicateBulkWrite 同族,调用方判分支不必再匹配文案。

没有叫 code,是因为契约在别处:ADR-0112 把 error.code 定为闭集(StandardErrorCode ∪ ERROR_CODE_LEDGER,都在 packages/spec/src/api/),而 rest-server.ts 会把抛出错误的 .code 直接提升到响应封套上。在这里加一个 .code,等于以副作用的方式凭空造出一个未注册的线上错误码 —— 正是这一族错误要消灭的 declared ≠ enforced 形状。真要上线,应当走 ledger 注册这条路(本轮 spec/** 零改动,不做)。

3. 检测:走解析后的 CEL AST,不再依赖 cel-js 的措辞

「这条条件是否引用了 previous」改由 collectCelRootIdentifiers(#4972 的构建闸门用的同一个工具)在 wrap 时算一次,旧的 fault 文本判断降为兜底。实测两条路今天答案一致,收益是这条面向作者的诊断不再挂在上游库的错误措辞上;顺带修掉一类会错判的形状:record.previous_status 是 record 下的成员名,AST 给出的 root 只有 record,不会被当成 previous 引用(按文本匹配就会)。工具已知的 caveat(comprehension 绑定变量会被算作 root)在这里是惰性的:该答案只在报错路径上被读,而 [1,2].exists(previous, previous > 1) 自绑变量、求值正常,永远走不到报错路径 —— 已就此加了用例。

4. ADR-0058 补遗 + anchors

按 #4889 / PR #4972 的先例:只追加,不改旧文。补遗记的是契约本身(批量写上 after 型 hook 与 record-change flow trigger 按行求值、按行触发,与 validation 侧 #3106 同源)、今天引擎的实测行为、rc 期以本诊断止血及其退役条件(#5038),以及 #5038 实现时必须一并定的代价(hook 触发 N 次、ctx.result 形状、onError 的按行语义、大批量上限)。scripts/adr-anchors.json 把 ADR-0058 + ADR-0112 锚到 hook-wrappers.ts。

回归测试

新增 packages/objectql/src/hook-condition-bulk-previous.test.ts(17 例),覆盖派单要求的四项:

(a) 批量写 + 引用 previous → limitation === 'bulk_write_previous_unbound'、文案含批量限制/当前版本/ADR-0058/[17.x] 批量写按行语义实现:hook 按行触发 + record-change trigger 按行绑定 previous/record(#4800/#4862 拍板 A) #5038/单记录出路;经真实引擎 engine.update(..., { multi: true }) 仍然写入失败(诊断不是豁免);delete 形状的批量写同样命中;
(b) 同一 hook 的单记录写入:previous 绑定、条件成立则 handler 运行、不成立则安静跳过、引擎路径上更新照常落库 —— 完全不变;
(c) 不引用 previous 的批量写:条件为真照跑、为假照跳、未声明键仍报「拼错」且不提批量;声明了但本次没写的字段保留自己的 bulk_write_stored_state_unavailable;
(d) 该场景下泛化 CEL 措辞不再作为结论抵达调用方(which this object does not declare / which is not bound for this operation 均不出现,原始 fault 仍作为事实保留在 fault 字段),外加 AST 路线的两条钉子(混合拼错时两半都报;record.previous_status 不误判)。

本地闸门(均在 flock /tmp/os-heavy-verify.lock 下、限堆、限包):

pnpm --filter @objectstack/objectql typecheck   → tsc --noEmit 无输出(通过)
pnpm --filter @objectstack/objectql test        → Test Files 114 passed, Tests 1810 passed
node scripts/check-adr-anchors.mjs              → OK (20 anchored file(s))
node scripts/check-error-code-casing.mjs        → ✓ no lowercase error codes (2756 files)
node scripts/check-nul-bytes.mjs / check-release-notes.mjs / check-doc-authoring.mjs → OK
npx eslint <改动文件> --no-inline-config        → 无输出

合并 origin/main(含 #5045 动了 packages/spec/src/contracts/objectql-engine.ts)后重装、重建依赖包并重跑 typecheck + 全量 objectql 测试,仍为 114/1810 全绿。

约束核对

packages/spec/** 与生成物:零改动;packages/metadata-protocol/src/protocol.ts:零改动;content/docs/releases/:未触碰。
改动面:packages/objectql/src/{hook-wrappers,index,core}.ts + 新测试、docs/adr/0058-*.md(仅追加补遗)、scripts/adr-anchors.json、一个 patch 级 changeset。engine.ts 未改 —— 批量分支的事实(input.id 缺失 + options.multi)已足够判定,不需要动那条同日被 fix(objectql,lint): 服务端补上 parent 作用域 readonlyWhen 的绑定与执行 (#4889) #4972/fix(objectql): resolve the NOW() defaultValue token in the engine (#4597) #4993/fix(objectql): 区分「同一实例的幂等重入」与「真正的 driver 名字冲突」(#4773) #5007 改过的路径。

已知的下游口径(留给 #5038,不在本单)

skills/objectstack-formula/SKILL.md §5 目前把「两条出路」写成 rewrite the condition without previous 或改单记录写入,并把「过渡条件只写在单记录写入上」讲成永久规则;#5038 的验收面已包含「文档与 formula skill 的 transition 写法口径随之收口」。本 PR 不动它:改 SKILL.md 会带出 gen:skill-docs 的生成物,与本轮「spec/生成物零改动」冲突。

🤖 Generated with Claude Code

https://claude.ai/code/session_01NrmBxj8rK2uGCnh9aipjwX

Generated by Claude Code

…author, and record the per-row contract (#5037) The rc-window stopgap half of the 2026-08-04 ruling on #4800 / #4862. A hook `condition` reading `previous` on a predicate (`multi: true`) write is unevaluable — the engine matches N rows, fires the hook once and binds no prior record — and since #4775 that rejects the write. #4861 already gave the case its own sentence, but that sentence predates the ruling and led with "rewrite the condition without `previous`", which is advice to silently turn a transition into a state test. The contract is now the opposite: on a bulk write, after hooks and record-change flow triggers evaluate and fire PER ROW. - `docs/adr/0058`: addendum recording the contract, today's measured behaviour, the stopgap and its expiry (#5038), and the consequences #5038 must price. - `hook-wrappers.ts`: the rejection states it is a CURRENT-VERSION limitation, cites the contract and #5038, leads with the single-record route, and prices the rewrite. `HookConditionError.limitation` (`bulk_write_previous_unbound` | `bulk_write_stored_state_unavailable`) is the machine-readable discriminator — deliberately not `code`, since ADR-0112 makes `error.code` a closed wire vocabulary and rest-server promotes a thrown error's `.code` onto the envelope. - "does this condition read `previous`" now comes from the parsed CEL AST (`collectCelRootIdentifiers`), computed once at wrap time, with the fault-text check as fallback — so the diagnosis no longer rides on cel-js's wording, and `record.previous_status` is correctly NOT a `previous` reference. - `scripts/adr-anchors.json`: anchors ADR-0058 + ADR-0112 to hook-wrappers.ts. Single-record writes, bulk writes whose conditions do not name `previous`, and the undeclared-key typo report are unchanged; fail loud takes no exception. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01NrmBxj8rK2uGCnh9aipjwX

vercel · 2026-08-04T01:13:47Z

The latest updates on your projects. Learn more about Vercel for GitHub.

1 Skipped Deployment

Project	Deployment	Actions	Updated (UTC)
objectstack	Ignored		Aug 4, 2026 1:13am

github-actions · 2026-08-04T01:14:06Z

📓 Docs Drift Check

This PR changes 1 package(s): @objectstack/objectql.

13 hand-written doc(s) reference the affected code and may need an implementation-accuracy re-verification:

content/docs/concepts/metadata-lifecycle.mdx (via @objectstack/objectql)
content/docs/data-modeling/formulas.mdx (via packages/objectql)
content/docs/deployment/migration-from-objectql.mdx (via @objectstack/objectql)
content/docs/deployment/vercel.mdx (via @objectstack/objectql)
content/docs/kernel/services-checklist.mdx (via @objectstack/objectql)
content/docs/kernel/services.mdx (via @objectstack/objectql)
content/docs/permissions/authentication.mdx (via @objectstack/objectql)
content/docs/plugins/index.mdx (via @objectstack/objectql)
content/docs/plugins/packages.mdx (via @objectstack/objectql)
content/docs/protocol/kernel/index.mdx (via @objectstack/objectql)
content/docs/protocol/objectql/query-syntax.mdx (via packages/objectql)
content/docs/protocol/objectql/state-machine.mdx (via @objectstack/objectql)
content/docs/releases/implementation-status.mdx (via @objectstack/objectql)

Advisory only. To re-verify, run the docs-accuracy-audit workflow scoped to these files:
node scripts/docs-audit/affected-docs.mjs origin/main → pass the list as args.docs.

github-actions Bot added size/l documentation Improvements or additions to documentation tests tooling labels Aug 4, 2026

claude Bot mentioned this pull request Aug 4, 2026

[rc 止血] 批量写上引用 previous 的 hook 条件:泛化失败换成点名批量限制的专门诊断 + ADR 补遗 #5037

Closed

xuyushun441-sys marked this pull request as ready for review August 4, 2026 01:27

xuyushun441-sys added this pull request to the merge queue Aug 4, 2026

Merged via the queue into main with commit 4335497 Aug 4, 2026
24 checks passed

xuyushun441-sys deleted the claude/issue-5037-bulk-previous-diagnostic branch August 4, 2026 01:34

claude Bot mentioned this pull request Aug 4, 2026

[PM] 分片分工登记表 — 谁在管哪个仓库的队列 #4604

Open

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

fix(objectql): 批量写上引用 previous 的 hook 条件 —— 点名「当前版本限制」的专门诊断 + ADR-0058 按行语义补遗 (#5037) - #5057

fix(objectql): 批量写上引用 previous 的 hook 条件 —— 点名「当前版本限制」的专门诊断 + ADR-0058 按行语义补遗 (#5037)#5057
xuyushun441-sys merged 1 commit into
mainfrom
claude/issue-5037-bulk-previous-diagnostic

xuyushun441-sys commented Aug 4, 2026

Uh oh!

vercel Bot commented Aug 4, 2026

Uh oh!

github-actions Bot commented Aug 4, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Uh oh!

Conversation

xuyushun441-sys commented Aug 4, 2026

先说清楚现状:第 1 项已部分落在 #4861 里

改了什么

1. 文案:从「换个写法」改成「这是当前版本的限制」

2. 判别位:limitation,并且故意不叫 code

3. 检测:走解析后的 CEL AST,不再依赖 cel-js 的措辞

4. ADR-0058 补遗 + anchors

回归测试

约束核对

已知的下游口径(留给 #5038,不在本单)

Uh oh!

vercel Bot commented Aug 4, 2026

Uh oh!

github-actions Bot commented Aug 4, 2026

📓 Docs Drift Check

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

2. 判别位:`limitation`,并且故意不叫 `code`