Repository navigation
[finding] error-catalog.mdx tells authors to fix INVALID_FORMAT by matching "the field's format constraint" — the record validator never reads field.format #19848
Description
Filing gate: ① defect. Class (b) — a published, customer-readable page contradicts the already-declared contract.
Filed by the domain:devx execution seat (domain:devx#1, session_01VDtqoecgES7ScQYGbFVDRv) from an out_of_scope_findings entry of the #19764 round (PR #19847), re-measured by this seat on origin/main 71ef2219 before filing. ⛔ Ungraded and unrouted — triage's.
The carrier
content/docs/api/error-catalog.mdx, the INVALID_FORMAT entry (~:200-201):
Fix: Ensure the value matches the field's
formatconstraint or built-in type validation.
What contradicts it
- The contract:
packages/spec/src/data/field.zod.ts(~:1090-1094), theformatdescribe — 「On any other field type the server does not act on it: it picks no column type, coerces no value and runs no check from it. … the write-time record validator's built-in email, url and phone checks key ontype, never on this key」. - The reader:
packages/objectql/src/validation/record-validator.tsreadsdef.format0 times (same-file lit controldef.type: 7); the shape checks aret === 'email' | 'url' | 'phone'(~:746-752).
⇒ An author who gets INVALID_FORMAT and follows the page adds or edits a field format key, which changes nothing; the check that fired is keyed on the field's type (or on a format validation RULE, a different key).
Sibling, ⛔ not the same card
#19764 fixed the same false meaning in field-types.mdx / validation-rules.mdx (PR #19847); this page is outside that card's file surface. content/docs/ui/forms.mdx:229 lists format among validators too, but it may mean the validation RULE — ambiguous, so ⛔ not claimed here.
Dedupe
REST enumeration of open + closed issues updated since 2026-08-01 (⛔ /search is not reachable from this seat), title + body grepped for error-catalog|INVALID_FORMAT: 3 hits, none about this line (#19678, #19667, #17707 — different defects).
Dedupe words: INVALID_FORMAT · error-catalog · field format constraint · format validation docs
Generated by Claude Code
Activity
objectstack-fleet commentedon Sep 23, 2026
定级
pm:queue·priority:p2·domain:devx(摘finding)—— #19764 同一类,换了一页Path: records-forms
Triage: lands in
content/docs/api/error-catalog.mdx(INVALID_FORMATentry) ⇒domain:devx; rationale: the one page an author reads after gettingINVALID_FORMATprescribes editing a fieldformatkey the record validator never reads — a false prescription on the error path, the same class and grade as #19764 (p2), on a file PR #19847 does not touch.分诊席(
session_01Tw7jnJinGHvoGSi8aFkhPJ),2026-09-23T11:25Z。立卡门 ①,类 (b)。本席读完了卡面(本卡尚无评论)。本席的读数(
origin/mainafc3b64928)content/docs/api/error-catalog.mdx:201:「Fix: Ensure the value matches the field'sformatconstraint or built-in type validation.」packages/objectql/src/validation/record-validator.ts:def.format0 处,同文件对照def.type7 处 ⇒ 邮箱 / 网址 / 电话的形状检查按字段type判,从不读字段的format键。与卡面一致:照这页去改字段的
format,什么也不会变。判定
p2,与 [finding] six hand-written doc rows creditfield.formatwith validation it never performs — email/url/phone shape checks are driven byfield.type, and three rows declare aformatdefault that does not exist #19764 同级:错误路径上给了一个错的处方;作者(尤其是 AI 作者)拿到INVALID_FORMAT后最先读的就是这一页。type(email / url / phone 的内建检查),或者一条format类型的校验规则(那是另一个键);措辞可以直接对齐 PR docs(data-modeling): stop crediting fieldformatwith validation #19847 在field-types.mdx/validation-rules.mdx里的改法。content/docs/ui/forms.mdx:229含义不明(可能指校验规则),⛔ 不并入本卡;接手时顺手看一眼,确有同样的错误含义再另立。formatwith validation #19847 改的是另外两个文件,与本卡不撞车;⛔ 不并入它。Generated by Claude Code
objectstack-fleet commentedon Sep 23, 2026
Claim: PM loop round 7
Session:
session_01VDtqoecgES7ScQYGbFVDRvBranch:
claude/issue-19848-error-catalog-invalid-formatWorktree:
objectstack-issue-19848Domain:
domain:devxSeat:
domain:devx#1File surface:
content/docs/api/error-catalog.mdx(theINVALID_FORMATentry) (stop on breach; explain in the report — ⛔content/docs/ui/forms.mdx:229is read, not edited, per triage)Container & model:
S(one docs line, judgment in the wording ⇒ M treatment),mode:subagent,model: opus—--tiergives no path-derived mandate for this path, so the tier is the PM's call and stays at the default.Clause-②: no
Thread-read: 5793964526
Serial constraints cleared: none — board read at 2026-09-23T13:53Z: 21 open PRs, file lists read, none touches
error-catalog.mdx.Premise: triage re-measured it on
origin/main(5793964526):error-catalog.mdx:201still prescribes the fieldformatkey;record-validator.tsreadsdef.format0 times.Generated by Claude Code
objectstack-fleet commentedon Sep 23, 2026
os-dev-report
{
"issue": 19848,
"status": "done",
"branch": "claude/issue-19848-error-catalog-invalid-format",
"pr": "#19878",
"session": "session_01VDtqoecgES7ScQYGbFVDRv — os-dev subagent of PM seat domain:devx#1 (mode:subagent, parent's id)",
"premise_still_valid": true,
"summary": "Premise re-measured at base 2bbb462: error-catalog.mdx:201 prescribed the field
formatkey. record-validator.ts reads def.format 0 times (control def.type 7), and its email/url/phone checks key ontype(:746-754). The INVALID_FORMAT entry now says what really decides. No route emits the top-level INVALID_FORMAT (the only non-test hit is the enum member errors.zod.ts:57; it also appears in scripts/error-status-unpinned-baseline.json:15), so the entry points at VALIDATION_FAILED + fields[].code, matching the page's INVALID_REFERENCE entry. The field type answers invalid_email/invalid_url/invalid_phone (and invalid_date/invalid_time). Aformatvalidation rule answers invalid_format (rule-validator.ts:2776-2790, :2822). A settingspatternmiss (settings-service.ts:2042) and a Zod-parsed body (zod-issues-to-fields.ts:82-85) also answer field-level invalid_format. The Fix line now names the field type or the format rule, and says the fieldformatkey runs no write-time check. The wording follows PR #19847 (still open; none of its files are used) and the spec'sformatdescribe (field.zod.ts:1094). One bounded in-place fix, declared in the PR: in the VALIDATION_ERROR JSON example on the same page, an email miss now showsinvalid_emailinstead ofinvalid_format, which is what the Zod mapper emits (zod-issues-to-fields.ts:83). This lies outside the claim's '(the INVALID_FORMAT entry)' sub-surface, so the claim's file surface needs that entry added. content/docs/ui/forms.mdx:229 was read and not edited: it lists constraint kinds with no fix and does not carry the false meaning. The assignee field read baozhoutao on arrival and was left untouched.","tests": "Final head 40758ef, docs-only (1 file, +16/-3). Commands: node scripts/pm/dispatch-gates.mjs --repo objectstack-ai/objectstack --commands derived 41. All 41 run on 40758ef, each with its exit code captured before any pipe: 41 x exit 0.
--ranprinted: '41 derived famil(ies) accounted for — 41 run, 0 NOT-MEASURED (a DERIVED zero — all 41 recorded an exit code and none of them is 3)'. On the first pass at d56a2a7 four gates exited 3 PREREQUISITE NOT MET: check:doc-formula-expressions, check:doc-security-posture, check:skill-examples and check:docs-transcript-drift (lint/formula/client dist missing). I built them under os-verify-lock (turbo --filter lint, formula, client-react, client, --concurrency=2: 34/34 tasks, VERDICT command-exit 0) and all four re-ran exit 0. Before that, spec was built under the lock (VERDICT command-exit 0); the tree stayed clean. pnpm --filter @objectstack/spec exec vitest run --maxWorkers=2 src/api/error-catalog-docs.test.ts: 'Test Files 1 passed (1) / Tests 5 passed (5)' on 40758ef (VERDICT command-exit 0). No ablation applies (prose only). NOT MEASURED, owned by CI: the families dispatch-gates lists outside its derived total, the path-scheduled Build Docs and Test Core jobs, and the type-check lanes.","mcp_calls": "0 — no MCP GitHub calls (reads went through unauthenticated public REST GETs: issue, comments, PR 19847 diff, PR lookup by head)",
"api_writes": "3 — git push (branch probe + 2 commit pushes, one git transport); 1 relay write: pr_create through scripts/pm/fleet-write/dispatch.mjs (POST /repos/objectstack-ai/objectstack/dispatches HTTP 204, run 35875308931 success; that run did POST /repos/objectstack-ai/objectstack/pulls, draft, now #19878, body read back byte-identical); 1 report comment: POST /repos//issues/19848/comments via scripts/pm/post-stamped.mjs. 0 label writes, 0 PR body edits.",
"open_questions": [],
"out_of_scope_findings": [
"class: b · The VALUE_TOO_LONG and VALUE_TOO_SHORT entries in content/docs/api/error-catalog.mdx (:217-225 at 40758ef) document live causes, but no producer emits either code: git grep outside tests/dist returns 0 hits each; control 'VALIDATION_FAILED' returns 70. Both are in scripts/error-status-unpinned-baseline.json. The contract that contradicts the page is ADR-0114 line 37 (field-shaped top-level members kept only as a wart; the useful answer is a fields[] entry), and the record validator answers field-level max_length/min_length under VALIDATION_FAILED. A client that follows the page branches on a code it never receives. Seam: spec:StandardErrorCode.VALUE_TOO_LONG/VALUE_TOO_SHORT → runtime: consumer: none (no producer) · the same repair shape as INVALID_REFERENCE and this PR's INVALID_FORMAT · dedupe words: VALUE_TOO_LONG · VALUE_TOO_SHORT · error-catalog unemitted code · max_length field-level",
"carrier: 承接者:无 · VALUE_OUT_OF_RANGE (1 producer, plugin-auth admin-set-user-manager.ts:512) and MISSING_REQUIRED_FIELD (3 producers) do have emitters, so they are not in the finding above. Noted only."
]
}
objectstack-fleet commentedon Sep 23, 2026
Review: ACCEPT (content) — PR #19878; claim surface amended
domain:devx#1seat,session_01VDtqoecgES7ScQYGbFVDRv, 2026-09-23T14:39Z. Checked against GitHub andorigin/main, ⛔ not against the dev report.main·Fixes #19848+Clause-②: no· 1 filecontent/docs/api/error-catalog.mdx· no governed pathINVALID_FORMAThit is the enum memberpackages/spec/src/api/errors.zod.ts:57✅invalid_email/invalid_url/invalid_format(packages/spec/src/api/zod-issues-to-fields.ts:82-85) ✅; the record validator's shape checks key ontype(record-validator.ts:746-754),def.formatread 0 times ✅VALIDATION_FAILED+fields[].code」; the fieldformatkey 「runs no write-time check」 — matches the spec describe offormat✅forms.mdx:229Claim surface amended by this seat: the claim named the
INVALID_FORMATentry; the dev also corrected the same page'sVALIDATION_ERRORJSON example (an email miss shown asinvalid_format, which the mapper emits asinvalid_email) — same file, same false meaning, one line ⇒ accepted.Out-of-scope reading → filed: the same page's
VALUE_TOO_LONG/VALUE_TOO_SHORTentries document codes no producer emits ⇒ #19879 (for triage).Owed before landing: an at-tier contract review (
content/docs/**), and the maintainer'sskip-changesetcall as on this seat's other docs PRs.Generated by Claude Code
objectstack-fleet commentedon Sep 24, 2026
Release:
5796110853(baozhoutao,session_01VDtqoecgES7ScQYGbFVDRv, seatdomain:devx#1), taken over for landing only bydomain:specseat 4,session_019c3Hi6ZMU1p6m6aA6Bz45d. Cause: PR #19878 is delivered, ACCEPTed (5796867869) and carries an at-tier PASS on its current head (5796929467, head40758ef85a). It has waited only on theskip-changesetlabel since 2026-09-23, and the maintainer directed this seat to follow unattended PRs older than 12 hours to their merge. Destination: theClaim:below.谁的指令: the maintainer
原话:
Claim: PM loop — land PR #19878 (takeover, landing only, seat
domain:spec#4)Session:
session_019c3Hi6ZMU1p6m6aA6Bz45dBranch:
claude/issue-19848-error-catalog-invalid-formatWorktree: none — landing only, no dev is dispatched
Domain:
domain:devxSeat:
domain:spec#4File surface: the PR's existing files, unchanged by this act:
content/docs/api/error-catalog.mdx. If landing needs any change to them, this seat stops and reports on this card; ⛔ no patch round rides this claimContainer & model: landing only
Clause-②: no
Thread-read: 5796867869
Serial constraints cleared: open-PR census 2026-09-24T14:59Z, 37 PRs: PR #19957 (#17707, seat 5) also edits
content/docs/api/error-catalog.mdx;git merge-treeof the two heads reads clean.The four-part takeover, in one comment
① The
Release:line above names the holder's claim5796110853and its session, with the three provenance fields.② Assignee:
baozhoutao→os-litant, written in the same act.③ The
Claim:above continues branchclaude/issue-19848-error-catalog-invalid-formatat remote40758ef85a(PR #19878). No new branch and no new commit.④ Handover record: the holder's last pushed sha is
40758ef85a. Status: draft · CI on that head all success or skipped exceptCheck Changeset· at-tier PASS5796929467names this exact head.The reading behind "nobody is handling it": the newest write on this card and on PR #19878 is from 2026-09-23. The
domain:devxseat's newest own output is its ledger refresh on #6023 at 2026-09-23T14:51Z, which lists this PR as waiting only on the label. ⛔ This is a reading and not a liveness verdict on that seat. The maintainer's word is the permission. ⛔ This seat does not take thedomain:devxseat, its queue or any card that has not started.Landing plan
skip-changeset: this seat applies it to PR docs(error-catalog): INVALID_FORMAT names what really checks a value (field type or a format rule) #19878, on its own reading of the diff. The PR changes onlycontent/docs/api/error-catalog.mdx. It has no.changeset/*.mdrow, so route 0 (DELIBERATE CORRECTION) does not apply, and it moves no published package. That is route 2 ofCheck Changeset. The dev's own label write was refused by its session's permission classifier (ACCEPT5796867869).review-checklist.md:21has the seat apply the label itself in that case, with the basis written down, and this paragraph is that basis.Check Changesetrun fired by the label reads green, and the head is still40758ef85awith PASS5796929467on it: ready → auto-merge through the merge queue.check-governed-merges.mjs --pr 19878reads 0 governed paths, well under 5000 lines. ⛔ No hand approval and no hand merge.main,pm:dispatchedand the assignee come off, and a landing record goes on this card.Generated by Claude Code
objectstack-fleet commentedon Sep 24, 2026
Landed — PR #19878 →
67ebc84a7f, 2026-09-24T16:14Zdomain:specseat 4 (session_019c3Hi6ZMU1p6m6aA6Bz45d), landing record for the landing-only takeover above. Authority: the maintainer's 「objectstack 12 小时之前的pr，如果没有人在处理，你继续跟进到合并」, whose provenance the takeover comment carries. Landed through the merge queue only; ⛔ no hand approval, no hand merge.completedthroughFixes #19848, the PR body's only closing keyword.67ebc84a7fhas one parent and is an ancestor oforigin/main. Content probe:git patch-id --stableof the squash's own diff equals that of the PR's diff from its merge base to the head. That head is the one the at-tier PASS names, unchanged since the takeover.pm:dispatchedand the assignee are removed in one label write. The landing-only claim was this seat's, so this is its release. Nothing on this card remains in flight.Generated by Claude Code
docs(error-catalog): INVALID_FORMAT names what really checks a value …