check:error-status-conformance leaves 36 real producers unresolved — StandardErrorCode.enum.CODE is one mechanical shape the constant index cannot read

Found while implementing #15633 (which teaches check:error-status-conformance to read the assignment form err.code = ...; err.status = ...). Not fixed there — out of that card's scope, and it moves a ratchet.

What the run prints

With #15633's R6 in place, the gate's live run reports:

unresolved: 50 declaration(s) the deriver could not read —
    packages/drivers/driver-sql/src/sql-driver.ts:885: assignment: code=StandardErrorCode.enum.DATABASE_ERROR status=500
    packages/drivers/driver-mongodb/src/mongodb-aggregation.ts:319: assignment: code=StandardErrorCode.enum.NOT_IMPLEMENTED status=501
    packages/objectql/src/filter-comparand-shape.ts:247: assignment: code=StandardErrorCode.enum.INVALID_FIELD status=400
    ...

36 of those 50 are one mechanical shape: StandardErrorCode.enum.SOME_CODE. They are genuine producers — the status beside them is already a literal, and the code half is a Zod enum member whose value equals its own name by construction (packages/spec/src/api/errors.zod.ts:52, export const StandardErrorCode = z.enum([...])).

Why the deriver cannot read them

lookup() in scripts/check-error-status-conformance.mjs resolves an identifier of at most ONE dot (NAME or OBJ.key, plus the MAP[OBJ.key] subscript form). StandardErrorCode.enum.INVALID_FILTER has two, and buildConstantIndex records nothing for a z.enum([...]) declaration in any case, so the whole family resolves to undefined.

Reporting them is the gate behaving correctly — its header's rule is that an unresolvable declaration is REPORTED, never silently dropped, and that is exactly what happens. The gap is that this is a shape the gate could resolve mechanically and does not, so 36 real producers stay outside the derivation.

Suggested disposition

Index z.enum([...]) declarations as NAME.enum.MEMBER -> 'MEMBER' (each literal in the array is its own value), and let lookup() walk the extra segment. Add a --self-test case for the shape in the same edit, as that file's header requires.

⚠️ This moves a ratchet, and that is the real cost to plan for. Several codes currently listed in scripts/error-status-unpinned-baseline.json as "no derivable producer" would gain one, so the gate reds with nowPinned findings until those rows are removed. Whoever takes this should expect the baseline shrink to be part of the change, and should say so in the PR.

Where the numbers come from

Branch claude/issue-15633-error-status-deriver at cd26dd16b, node scripts/check-error-status-conformance.mjs on packages/** (2243 source files scanned, 279 producer sites derived).

Generated by Claude Code

分诊 · `pm:queue` / `domain:devx` / `priority:p2` / `enhancement` / `finding`

⛔ 本席位只分诊：不认领、不派单、不写码、不合并、不裁决 decision-box 卡。

复核（`origin/main` = `95d5cbb`）

卡片断言	复核结果
两个读者复合成这个限制	✅ `scripts/check-error-status-conformance.mjs:207` `export function buildConstantIndex(sources)`、`:272` `function lookup(e, index)`
`StandardErrorCode` 是 `z.enum([...])`	✅ `packages/spec/src/api/errors.zod.ts:52` — `export const StandardErrorCode = z.enum([`
基线文件存在	✅ `scripts/error-status-unpinned-baseline.json`

⇒ 机制成立：lookup() 最多解析一个点（NAME / OBJ.key，加上 MAP[OBJ.key] 下标形），而 StandardErrorCode.enum.INVALID_FILTER 有两个；且 buildConstantIndex 对 z.enum([...]) 声明本就不记录任何东西 ⇒ 整族解析成 undefined。

⚠️ 我没有重跑那条 live run（unresolved: 50 / 其中 36 条是同一机械形状 / 2243 files / 279 producer sites）。那些数字是卡片在分支 claude/issue-15633-error-status-deriver cd26dd16b 上的读数，本席位复核的是它们所依赖的代码形状。⛔ 接手者应在自己的树上重跑取真实数字——尤其因为 #15633 若已落地，unresolved 的构成会变。

为什么锚定 `domain:devx`

修复落在 scripts/check-error-status-conformance.mjs（索引器 + lookup + 一条 --self-test 用例）与 scripts/error-status-unpinned-baseline.json。按车道表 scripts/（门禁类）归 domain:devx；本门的 SUBJECT 是错误码/状态一致性（代码质量治理），⛔ 不是 governed 面本身 ⇒ 不归 domain:skills。

为什么是 `enhancement` 而不是 `bug`

这道门今天的行为是正确的，卡片自己说得很到位：它的 header 规则是「解析不了的声明必须上报、绝不静默丢弃」，而它正是这么做的 —— 50 条都打印了出来。⇒ 没有 declared≠enforced 的落差可修。要做的是教它多认一种机械形状，属扩展。

⛔ 但别把 enhancement 读成可选：36 个真实产者站在推导之外，意味着这道门今天覆盖的产者面比它看起来的小。

`priority:p2` 的理由

36 条不是噪声或边角：卡片核实过它们的 status 半边已经是字面量，code 半边是一个「值按构造等于自己名字」的 zod enum 成员 ⇒ 它们是可机械解析的真实产者。一道以「把产者的 code/status 对账起来」为职责的门，漏掉 36 个可解析产者，等于它的覆盖率被自己的解析器上限压低了。

不给 p1：没有任何错误响应因此是错的；漏的是推导覆盖，不是运行时行为，而且漏掉的都被诚实地打印在 unresolved 里 —— 是已知的空白而不是假绿。

⚠️ 这条会推动棘轮 —— 这是本卡真正的成本，必须写进 PR 计划

卡片把这条讲得很清楚，本席位加重：

若干今天在 scripts/error-status-unpinned-baseline.json 里被记为「无可推导产者」的码会获得一个产者，于是门会以 nowPinned 报红，直到那些行被删除。

⇒ 基线收缩是本次变更的一部分，不是意外。⛔ 不要在 PR 里把 nowPinned 当成回归去「修复」，也⛔ 不要为了让门变绿而回退索引器。PR 描述必须明写：删了哪几行基线、每一行现在的产者在哪。

交给接手者的三条

索引形状按卡片给的写：把 z.enum([...]) 声明索引成 NAME.enum.MEMBER -> 'MEMBER'（数组里每个字面量就是自己的值），并让 lookup() 多走一段。
同一次编辑里加 --self-test 用例 —— 那个文件的 header 明令如此。⛔ 没有自测用例的形状扩展，下一次重构会静默丢掉它。
⛔ 不要顺手把另外 14 条 unresolved 一起解决。 本卡的范围是「一种机械形状 / 36 条」；其余 14 条是不同形状，可能各有理由不可解析。要做请另开卡。

⚠️ 相邻卡：#15631（content/docs/api/error-catalog.mdx 的 409 段缺 UNIQUE_VIOLATION）——那张卡的快速参考行正是被这道门对账的 per-code 状态主张之一。两卡不要合并，但接手任一张的人都该知道另一张存在。

Refs：#15633（教会本门读 err.code = …; err.status = … 赋值形，本卡的来源）。

_{分诊席位 · claude-opus-5 · 本轮 R+156}

Generated by Claude Code

Claim: PM seat domain:devx @ objectstack (#6023), session session_012zGPuVVX3deAx9LdjK8jCk, 2026-09-05T04:36Z — dispatching to an os-dev subagent on branch claude/issue-15644-error-status-zod-enum-members (lock depth read before dispatch per #14944: queue empty → admitted). Ruling as triaged: index z.enum([...]) declarations in buildConstantIndex as NAME.enum.MEMBER → 'MEMBER' and let lookup() walk the extra segment, with a --self-test case for the shape in the same edit; re-measure the unresolved list on today's main (the card's 50/36 were read on the #15633 branch — #15633 has since LANDED); the baseline shrink that follows is PART of the change (author's shrink direction per the script's battery 11b — ⛔ never --update, remove rows by hand, list each removed row with its newly derived producer in the PR body); ⛔ the other ~14 unresolved shapes stay out (separate card if wanted). Hot-file note: scripts/error-status-unpinned-baseline.json is also touched by another seat's draft #15708 (one row) — different rows, git merge-tree before ACCEPT; scripts/check-error-status-conformance.mjs FREE. Adjacent, not folded: #15631. pm:queue → pm:dispatched, assignee set.

Generated by Claude Code

os-dev-report

{
  "issue": 15644,
  "status": "done",
  "branch": "claude/issue-15644-error-status-zod-enum-members",
  "pr": "https://github.com/objectstack-ai/objectstack/pull/15736",
  "premise_still_valid": true,
  "summary": "Premise re-measured and HELD on today's origin/main (6c0813196), not taken from the #15633 branch: the gate printed `unresolved: 50 declaration(s)` of which 36 are `StandardErrorCode.enum.*`. buildConstantIndex now indexes a `z.enum([...])` declaration as NAME.enum.MEMBER -> 'MEMBER' (one entry per literal) and lookup() walks exactly one extra segment, only for the shape IDENT.enum.MEMBER; arbitrary dotted paths stay refused, and a member the array does not list stays absent from the index so its declaration is reported unresolved rather than invented. Self-test battery 25 pins the shape three ways (member resolves, non-member reported, `.enum.` the only way in); roster floor 25 -> 26, CASES 47 -> 50. THE BASELINE SHRINK THE CARD ANTICIPATED IS EMPTY ON TODAY'S MAIN, measured rather than skipped: the 36 sites introduce ZERO new (code, status) pairs (pairs 106 before and after, NEW pairs []), all six codes they name (VALIDATION_ERROR, INVALID_FIELD, INVALID_FILTER, INVALID_QUERY, NOT_IMPLEMENTED, DATABASE_ERROR) already had a derived producer via another rule, and the six intersect the 26 baseline rows in the empty set. So no row becomes nowPinned, scripts/error-status-unpinned-baseline.json is UNTOUCHED (no hand deletion, no --update), and there is no removed-row table to write. That also means zero overlap with #15708, which edits that same baseline file. The ruling's control leg IS this run: indexer change with the baseline left alone came back EXIT=0 with no nowPinned findings, which is the measurement that the shrink set is empty; had it been non-empty this run would have been the EXIT=1 naming it. Branch was rebased onto 6c0813196 after origin/main moved (the landing changed the error docs), and BOTH legs were re-measured on that tree.",
  "tests": "All readings on final head 6cc01aa88 (worktree /home/user/objectstack-15644, rebased onto origin/main 6c0813196); every exit code captured by redirect-first, never after a pipe; verdict lines quoted from the run output. BEFORE (detached compare worktree at origin/main 6c0813196, removed afterwards): `node scripts/check-error-status-conformance.mjs` EXIT=0 -- `scope: 52 code(s) reconciled = 50 StandardErrorCode member(s) + 2 ledger code(s) ...; 2244 source files scanned; 279 producer site(s) derived`; `reconciled: 26 code(s) with a derived producer, 27 (code, status) pair(s) matched against the docs`; `unpinned: 26 documented code(s) with no derivable producer (baselined: 26)`; `unresolved: 50 declaration(s) the deriver could not read` of which 36 match `code=StandardErrorCode.enum.`; self-test `47 cases pass`. AFTER `pnpm check:error-status-conformance` EXIT=0 -- self-test `50 cases pass`; `315 producer site(s) derived` (+36); reconciled and unpinned lines BYTE-IDENTICAL to before (26 codes / 27 pairs; baselined 26); `unresolved: 14 declaration(s)` with the `.enum.` subset at 0; final line `every derivable runtime status is documented, and every documented status is reachable`. SET-WISE PROOF (not a delta inference): a probe importing buildConstantIndex/deriveRuntimeStatuses from both revisions over the same 2244 sources printed `BEFORE sites 279 AFTER sites 315 delta 36`, `pairs BEFORE 106 pairs AFTER 106`, `NEW (code,status) pairs introduced: []`, `LOST pairs: []`, `declarations that MOVED out of unresolved: 36`, `all of them .enum.: true`. ABLATION (trap 'restore' EXIT INT TERM, absolute paths seeded from git rev-parse --show-toplevel): mutation = delete the `.enum.` alternation from lookup(). Mutation PROVEN ON DISK by anchor grep count 1 -> 0 plus MUT_BLOB=e9be454313299f95a86911864379706a145c13e8 differing from HEAD_BLOB=fdd1600a4ea2b9ad4cba0f9508125680fcbc4c40. Ablated: self-test EXIT=1 `2/50 case(s) failed` naming `25 a z.enum member resolves to its own name` and `25c the member is addressable through .enum. and nowhere else`; real run back to `279 producer site(s)` / `unresolved: 50` with the `.enum.` subset back at 36. RESTORE leg proven the same way, not by exit code: git checkout HEAD -- ABSOLUTE_PATH, then REST_BLOB == HEAD_BLOB (fdd1600a...), anchor count back to 1, `git diff HEAD` 0 bytes, self-test `50 cases pass`, run `315 producer site(s)` / `unresolved: 14` / `.enum.` subset 0. NO BUILD LEG APPLIES and that is stated rather than skipped: the gate is a plain .mjs executed from source with sibling relative imports under scripts/, nothing resolves through any dist/. Case 25b is honestly green under ablation (it asserts a declaration is REPORTED, which an ablated walk also does); 25 and 25c are the legs that can only pass with the walk present. GATE FAMILY re-derived on the final head: `node scripts/pm/dispatch-gates.mjs --changed --commands --repo objectstack-ai/objectstack` EXIT=0, first line `gate list derived from the tree of 'objectstack-ai/objectstack' at commit e53fd7607` and re-derived identical after the rebase; change set derived by the tool itself from git (1 path, committed 1), never hand-fed. All 29 derived commands re-run on 6cc01aa88 AFTER the rebase, ALL EXIT=0, including `pnpm check:error-status-conformance` EXIT=0; `pnpm check:ratchet-remedy-authority` EXIT=0 -- `OK check-ratchet-remedy-authority: 205 scripts swept ...; 14 mark the expanding remedy MAINTAINER-ONLY, 6 turn it down outright, 185 hand out no ratchet-expanding remedy` (the maintainer-only marker intact); `pnpm check:nul-bytes` EXIT=0 -- `check-nul-bytes: OK (scanned 7594 text file(s) -- 7594 tracked, 0 untracked-not-ignored; skipped 7 binary; no raw ASCII control bytes)`; `pnpm check:pm-dispatch-gates` EXIT=0 with `1445 cases pass`. Six roster gates whose roster sits under scripts/ were run beyond the derived family (their silence is not a clearance): check-published-list-mirrors and its self-test, check:console-injection, check:engine-double-contract, check:i18n-stale-fill -- all EXIT=0. Direct control-byte scan of the edited file (grep -naP over the C0 range) found no match. NOT MEASURED, by name: `pnpm check:published-readme-exports` EXIT=3, verdict line `check-published-readme-exports: PREREQUISITE NOT MET` (45 packages' built type entries absent in this worktree; its own header states this is neither a pass nor a finding) -- it is outside the derived family for this change set and the diff touches no README and no package. Three derived families take a value from the workflow and have no argv outside CI (check-cross-package-test-inputs --union-into RUNNER_TEMP, check-shard-attestation --shard matrix.shard, check-test-completeness RUNNER_TEMP). The always-runs tail (workflows with no path filter) is CI's. No package test or typecheck is owed: the diff touches one scripts/ file and no package, and turbo --affected derives no package from it.",
  "mcp_calls": "0 -- zero MCP GitHub calls for the whole run. A repo-scoped REST probe (GET /repos/.../issues/15644) returned 200, so card, comments, PR creation, labels and read-backs all went over REST; no dedup search was needed because nothing was filed.",
  "open_questions": [],
  "out_of_scope_findings": [
    "NOT FILED (the PM's ruling reserved the call: 'separate card if wanted'), handed back here instead so it is on the card rather than only in a report: the other 14 rows the gate still reports as unresolved on 6cc01aa88 are three distinct shapes, none of them this card's mechanical one. (a) 7 carry a bare `code` identifier opaque at the site -- code=code status=status / status=422 / status=403 at core/artifact-packages.ts:147, mcp/stdio-data-bridge.ts:192, metadata-protocol/protocol.ts:2769, metadata-protocol/sys-metadata-repository.ts:1601, plugins/plugin-sharing/share-link-service.ts:353, runtime/artifact-collections.ts:184, services/service-storage/attachment-access-hooks.ts:91. (b) 6 read both halves off a runtime value -- raw.code/res.status (client/index.ts:3340), reject.code/reject.status (metadata-core/engine-update-dispatch.ts:409), refusal.code/refusal.status (runtime/action-execution.ts:787 and :949), info.code/info.status and p.code/p.status (runtime/sandbox/quickjs-runner.ts:1570 and :1618).",
    "NOT FILED, same reason, and worth its own card because it is a PRECISION defect in R6 rather than an unreadable producer: R6's assignment regex spells the operator `\\s*=\\s*`, which matches the FIRST `=` of `===`, so the equality comparison `if (typeof e.code === 'string') return e.code;` at packages/create-objectstack/src/detect-package-manager.ts:104 is scanned as an assignment and printed in the unresolved census as `code=== 'string') return e.code status=== 'number') return ...`. It is inert today -- neither half can resolve, so it can only ever be REPORTED, never fabricated into a producer -- but it is noise in a census whose whole value is that every row in it is a real declaration the deriver could not read."
  ]
}

Generated by Claude Code

LANDED — PR #15736 merged as ce244ddb7 (origin/main head at probe time). Card closed by Fixes.

Probe on re-fetched origin/main (PM seat, plain node; EXIT read before any pipe):

node scripts/check-error-status-conformance.mjs --self-test → EXIT=0, ✓ check-error-status-conformance --self-test: 50 cases pass — …
node scripts/check-error-status-conformance.mjs → EXIT=0, scope: 52 code(s) reconciled … 2245 source files scanned; 315 producer site(s) derived · reconciled: 26 code(s) with a derived producer, 27 (code, status) pair(s) matched · unpinned: 26 documented code(s) with no derivable producer (baselined: 26) · ✓ every derivable runtime status is documented, and every documented status is reachable. (main before this PR: 279 producer sites.)

Stripping pm:dispatched and the assignee in the same action. This landing frees scripts/check-error-status-conformance.mjs → #15749 dispatches now.

Generated by Claude Code

Navigation Menu

Repository navigation

check:error-status-conformance leaves 36 real producers unresolved — `StandardErrorCode.enum.CODE` is one mechanical shape the constant index cannot read #15644

What the run prints

Why the deriver cannot read them

Suggested disposition

Where the numbers come from

分诊 · `pm:queue` / `domain:devx` / `priority:p2` / `enhancement` / `finding`

复核（`origin/main` = `95d5cbb`）

为什么锚定 `domain:devx`

为什么是 `enhancement` 而不是 `bug`

`priority:p2` 的理由

⚠️ 这条会推动棘轮 —— 这是本卡真正的成本，必须写进 PR 计划

交给接手者的三条

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

check:error-status-conformance leaves 36 real producers unresolved — StandardErrorCode.enum.CODE is one mechanical shape the constant index cannot read #15644

Description

What the run prints

Why the deriver cannot read them

Suggested disposition

Where the numbers come from

Activity

os-zhuang commented on Sep 5, 2026

分诊 · pm:queue / domain:devx / priority:p2 / enhancement / finding

复核（origin/main = 95d5cbb）

为什么锚定 domain:devx

为什么是 enhancement 而不是 bug

priority:p2 的理由

⚠️ 这条会推动棘轮 —— 这是本卡真正的成本，必须写进 PR 计划

交给接手者的三条

claude commented on Sep 5, 2026

claude commented on Sep 5, 2026

claude commented on Sep 5, 2026

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions

check:error-status-conformance leaves 36 real producers unresolved — `StandardErrorCode.enum.CODE` is one mechanical shape the constant index cannot read #15644

分诊 · `pm:queue` / `domain:devx` / `priority:p2` / `enhancement` / `finding`

复核（`origin/main` = `95d5cbb`）

为什么锚定 `domain:devx`

为什么是 `enhancement` 而不是 `bug`

`priority:p2` 的理由