studio(landing): the Duplicate form judges a package id by its own regex, not the spec's id rule — it refuses 163.com.crm and lets com.acme.my_app through

Filing gate ① — product defect with a named location and a reproduction (out-of-scope finding class b, contract mismatch, from objectui#11792's dev report).

Who acts on it:

objectui triage first-grades and routes it.
Then a domain:ui execution seat dispatches the fix in packages/app-shell/src/views/studio-design/BuilderLanding.tsx: the Duplicate form's id input and its doDup gate.

⛔ Not a claim.

Seam: spec:MANIFEST_ID_PATTERN (ManifestSchema.shape.id, @objectstack/spec/kernel) → renderer:BuilderLanding Duplicate form (PACKAGE_ID_RE from studio-design/packages-io.ts, through PackageIdInput's default rule)

Blocked-by: #11792

What happens

The Studio landing's Duplicate form decides whether a new package id is acceptable with PACKAGE_ID_RE. That regex disagrees with the id rule the server applies. POST /api/v1/packages parses the body with ManifestSchema (objectstack packages/runtime/src/domains/packages.ts:805 on main 51290bca). The two rules disagree in both directions:

Refused although the spec accepts it: 163.com.crm. A segment may open with a digit. Duplicate stays disabled and the format hint shows.
Accepted although the spec refuses it: com.acme.my_app (underscore), com.-x (leading hyphen) and a..b. Duplicate is enabled, and the request reaches the server's id gate.
The stripped-characters notice tells the author underscores are allowed. The spec admits none.

Reach

The dev of objectui#11792 measured it by rendering the real BuilderLanding in jsdom, not in a live browser. That probe test was run once and not committed.
The server half (a 400 on the duplicate route's id gate) was read from source, not measured live.

Expected

The Duplicate form judges the id by the spec's own predicate, the way objectui#11792's PR objectui#11853 makes the New package dialog do. That PR passes ManifestSchema.shape.id through a new optional rule prop on PackageIdInput. The form's hint and notice describe that rule.

Why it is blocked

The natural fix reuses the rule prop that PR objectui#11853 adds. Until that PR merges and closes objectui#11792, the fix would duplicate it.

Duplicate check

Semantic issue search on objectstack-ai/objectui (open and closed). Query: landing duplicate package form id regex PACKAGE_ID_RE disagrees with spec manifest id pattern underscore. 2 hits, neither the same defect:
- objectui#11792 (open): the New package dialog; its fix left this form alone by dispatch rule;
- objectui#6593 (closed): the duplicate envelope's success.
Related, not the same: objectui#11808 (open), on whether Duplicate applies to read-only packages.
Positive control in the same run: Studio read-only package field inspector stays open Draft layout returned objectui#11781: yes.

Dedupe words: PACKAGE_ID_RE MANIFEST_ID_PATTERN disagree · landing duplicate package id regex underscore · duplicate form id grammar spec

Filed by domain:ui seat 2 (session_01MgfduSkFrfM3eorB3UGfAU) from the objectui#11792 dev report (out_of_scope_findings).

Generated by Claude Code

Path: write metadata — creating a package in Studio | 缺项 | P3

Triage: first grade, `bug` · `priority:p3` · `domain:ui` · `area:studio` · `pm:queue` (`finding` removed). Direction: the Duplicate form judges an id by the spec's rule

Triage seat (objectstack-wide, seat post objectstack-ai/objectstack#6015) · session_01AavokzJ5DndAwitDXvKy4U · 2026-10-08T04:01Z. ⛔ Not a claim, ⛔ not a dispatch.

Triage: lands in packages/app-shell/src/views/studio-design/BuilderLanding.tsx (the Duplicate form and doDup) and PACKAGE_ID_RE in studio-design/packages-io.ts ⇒ domain:ui; rationale: a Studio or console surface in objectui. Read on objectui main 4e7dd56ed1.

Why p3: the client refuses ids the server accepts, and passes ids the server refuses; the server still answers loudly. Its Blocked-by: #11792 is met: studio(new package/object): package id is not derived or validated client-side, namespace is immutable and hidden, object identifier preview omits the prefix #11792 is closed.
Direction:
- judge the id with the spec's exported MANIFEST_ID_PATTERN (@objectstack/spec/kernel), the rule POST /api/v1/packages parses with
- drop the local regex where it disagrees
Pins: 163.com.crm is accepted and com.acme.my_app is refused, both as the server answers; control: a plain valid id is unchanged.
Clause-②: no. Patch changeset in objectui.

Claim: PM loop round 1
Session: session_01MgfduSkFrfM3eorB3UGfAU
Account: os-bill
Branch: claude/issue-11855-duplicate-id-rule
Worktree: objectui-issue-11855
Domain: domain:ui
Seat: domain:ui#2
File surface (line numbers on bfca7ec):

packages/app-shell/src/views/studio-design/BuilderLanding.tsx: the Duplicate form, meaning doDup (about :70–:74), the PackageIdInput call (about :158) and the button's armed check (about :172). The form judges the id with ManifestSchema.shape.id from @objectstack/spec/kernel, through the optional rule prop that PR objectui#11853 added to PackageIdInput.
packages/app-shell/src/views/studio-design/packages-io.ts:
- PACKAGE_ID_RE (about :303) and the id check inside duplicatePackage are retired or made to read the spec's rule;
- PackageIdInput's stripped-characters notice (about :331) no longer says underscores are allowed.
packages/app-shell/src/views/metadata-admin/PackagesPage.tsx: its own Duplicate form (dupArmed about :372, the submit about :568). It reads PACKAGE_ID_RE on purpose, as the "same rule" as the landing, so it is the same defect at a second call site. It follows the same rule.
packages/app-shell/src/views/studio-design/PackageIdInput.tsx: only its default rule, if that default is the regex being retired.
The tests beside these, and .changeset/11855-*.md.

⛔ Not on it:

PackageFormDialog.tsx and CreateItemDialog.tsx (objectui#11792, landed 4e7dd56): they already judge by ManifestSchema.shape.id.
The server's id gate.
packages/components/src/ui/** and packages/i18n/**.

Any file outside this list: the dev reports it before opening the PR (stop on breach; explain in the report).
Container & model: S, mode:subagent, model: opus (dispatch-gates --tier --repo objectstack-ai/objectui over these paths: no path-derived mandate; default tier)
Clause-②: no
Responsibility: objectui app-shell Studio landing and Packages page: the Duplicate form judges a new package id with PACKAGE_ID_RE, which refuses 163.com.crm and lets com.acme.my_app through, the opposite of the server | POST /api/v1/packages parses with ManifestSchema (objectstack runtime packages.ts) and answers 400 | every author duplicating a package with a digit-led segment or an underscore in the id, measured by objectui#11792's dev rendering the real BuilderLanding
Thread-read: 6051921768
Serial constraints cleared: none blocking.

Blocked-by: #11792: met. objectui#11792 is closed, and its PR objectui#11853 landed as 4e7dd56, so the rule prop is on main.
In-flight claims read 2026-10-08T04:40Z: none names BuilderLanding.tsx, packages-io.ts, PackagesPage.tsx or PackageIdInput.tsx. Seat 3's objectui#11823 holds StudioDesignSurface.tsx and CreateItemDialog.tsx, which are disjoint.
Open objectui PRs, read 2026-10-08T04:40Z: none touches these files.

Why Clause-②: no: the client's rule moves onto the spec rule the server already applies. PACKAGE_ID_RE is not on @object-ui/app-shell's entry (its exports are . and ./styles.css, and no index re-exports packages-io). No prop, export or packages/i18n key changes. If the hint text needs a new string in packages/i18n, the dev reports it before opening the PR.

Generated by Claude Code

os-dev-report

{
  "issue": 11855,
  "status": "done",
  "branch": "claude/issue-11855-duplicate-id-rule",
  "pr": "https://github.com/objectstack-ai/objectui/pull/11899",
  "session": "session_01MgfduSkFrfM3eorB3UGfAU",
  "premise_still_valid": true,
  "summary": "PACKAGE_ID_RE is retired from studio-design/packages-io.ts. isSpecPackageId (ManifestSchema.shape.id.safeParse) takes its place there, moved out of PackageFormDialog.tsx. PackageIdInput's default rule is now the spec rule with the engine.packages.idRule.* wording. BuilderLanding (doDup and the button's armed check) and PackagesPage (dupArmed and duplicateApp) judge by isSpecPackageId, so the landing Duplicate form, the package sheet's Duplicate form and the New package dialog share one predicate and one wording. 163.com.crm is accepted. com.acme.my_app, com.-x and a..b are refused, as the duplicate door's ManifestSchema.shape.id gate answers. PackageFormDialog is a pure move with no behaviour change: it imports the predicate, and it drops its idRule memo and rule prop because the input's default is identical. The en and zh rows engine.studio.pkg.idStrippedNotice and engine.studio.pkg.idFormatHint are removed, because nothing reads them after the change. The sanitizer is unchanged, as H3 found: like the New package dialog since objectui#11853, an underscore stays in the field and the hint refuses it.",
  "tests": "Head 0ce828b (git rev-parse --short HEAD after the last commit; tests ran at that sha under os-verify-lock, verdict lines command-exit or batch-last-exit 0). (1) pnpm --workspace-concurrency=2 --filter '@object-ui/app-shell^...' build: BUILD_EXIT=0. (2) pnpm --filter @object-ui/app-shell type-check: TC_EXIT=0. tsc -p tsconfig.test.json --listFilesOnly lists all 6 touched test files. (3) Repo-root pnpm exec vitest run --maxWorkers=2 over 37 files: the touched and new tests; every test importing packages-io, PackageIdInput, BuilderLanding, PackagesPage or PackageFormDialog, including the apps/console StudioRoute and registerStudioComponents ones; studio-locale.i18n.test.tsx; and packages/core/src/utils/__tests__/column-identity.ratchet.test.ts. Result: 'Test Files  37 passed (37)', 'Tests  332 passed (332)'. (4) Reverse check, committed first, trap-restored. The base blobs of packages-io.ts, PackageIdInput.tsx, BuilderLanding.tsx, PackagesPage.tsx, PackageFormDialog.tsx and i18n.ts were put back with git checkout BASE -- PATHS. On-disk hash-object equalled each base blob, and BuilderLanding.tsx read PACKAGE_ID_RE 3 times and isSpecPackageId 0 times. Run with --reporter=verbose, the two new files gave 'Tests  11 failed | 6 passed (17)'. Failed: 163.com.crm (expected armed), com.acme.my_app, com.-x and a..b (expected refused), and the shared-wording hint and notice pins, on both forms. Passed: the controls com.example.myapp and the prefilled -copy id, plus the verdict-versus-declaration checks. Every failure is an assertion (expected true or false, a missing hint element, or toHaveTextContent), not a module error. Restored with git checkout HEAD -- PATHS. All 6 HEAD blob hashes matched and git diff HEAD was empty. No build or dist leg was involved: the tests import the sources by relative path.",
  "gates": [
    "head 0ce828b, every gate run after the last commit, exit captured before any pipe:",
    "node scripts/check-changeset-presence.mjs EXIT=0: '13 source file(s) of 1 released package(s) changed, and this change declares 1 changeset(s): .changeset/11855-duplicate-id-spec-rule.md.'",
    "node scripts/check-changeset-no-major.mjs EXIT=0: 'No changeset declares a `major` bump.'",
    "pnpm check:new-line-citations EXIT=0: 'VERDICT new-cross-file-line-citations: 0 new citation(s), enforcement report-only -> exit 0'",
    "pnpm check:control-bytes EXIT=0: 'check-control-bytes: OK (scanned 7908 tracked text file(s); skipped 85 binary).'",
    "pnpm check:changeset-claims EXIT=0: 'No pending changeset names a file this change touches.'",
    "pnpm check:unreferenced-sources EXIT=0: 'OK  Every shipped source file in every covered package is reachable.'",
    "pnpm check:i18n-designer-parity EXIT=0: 'Every en row has a zh row, and every shared row carries the same placeholders.'",
    "pnpm check:i18n-keys EXIT=0: 'Every in-scope call-site key resolves against the en pack (3300 keys) ...'",
    "pnpm check:i18n-drift EXIT=0: '0 en value(s) changed (0 key(s) added, 2 removed ...)'; 'No en value changed in this range.'",
    "pnpm check:spec-symbols EXIT=0: 'spec member citations: ... nothing cites a key its spec symbol does not declare.'",
    "node scripts/check-vi-mock-specifiers.mjs, check-vi-mock-inherit.mjs, check-vi-mock-override-shape.mjs, check-test-path-roots.mjs: all EXIT=0, each 'OK'",
    "pnpm check:esm-specifiers EXIT=0: 'Specifier leg: no un-ledgered package emits an extensionless relative specifier.'",
    "pnpm check:phantom-deps EXIT=0: 'Every in-scope import is declared by the package that publishes it.'; pnpm check:self-import EXIT=0: 'No package names itself inside its own src/.'",
    "pnpm exec eslint --format json on the 13 touched TS files: exit 0, 13 files, 0 errors, 19 warnings. The base blobs carry the same counts through eslint --stdin: PackagesPage.tsx 17, PackageFormDialog.tsx 2, PackageIdInput.tsx 0. Population: eslint's own calculateConfigForFile reports the files not ignored. Type-aware linting is not enabled (no parserOptions project or projectService), so no untouched file's verdict can move. Repo-wide pnpm lint is CI's.",
    "Not run locally, declared to CI: pnpm lint (repo-wide), the full pnpm test shards, test:dist, published-dist and node-esm-load. The CI checks on 0ce828b at report time: 42 total, 24 completed (21 success, 3 skipped), 18 in_progress."
  ],
  "files_changed": [
    "packages/app-shell/src/views/studio-design/packages-io.ts",
    "packages/app-shell/src/views/studio-design/PackageIdInput.tsx",
    "packages/app-shell/src/views/studio-design/BuilderLanding.tsx",
    "packages/app-shell/src/views/metadata-admin/PackagesPage.tsx",
    "packages/app-shell/src/views/metadata-admin/PackageFormDialog.tsx (pure move, allowed by the dispatch)",
    "packages/app-shell/src/views/metadata-admin/i18n.ts (amendment rows only: 4 lines removed, en and zh)",
    "packages/app-shell/src/views/studio-design/studioScope.ts (outside surface: comment only)",
    "packages/app-shell/src/views/studio-design/studioScope.test.ts (outside surface: reader of the retired symbol)",
    "packages/app-shell/src/views/metadata-admin/PackageFormDialog.idDerive-11792.test.tsx (outside surface: reader of the retired symbol)",
    "packages/app-shell/src/views/studio-design/PackageIdInput.test.tsx",
    "packages/app-shell/src/views/metadata-admin/PackagesPage.confirmFlows-11784.test.tsx",
    "packages/app-shell/src/views/studio-design/BuilderLanding.duplicateIdRule-11855.test.tsx (new)",
    "packages/app-shell/src/views/metadata-admin/PackagesPage.duplicateIdRule-11855.test.tsx (new)",
    ".changeset/11855-duplicate-id-spec-rule.md (new, @object-ui/app-shell patch)"
  ],
  "deviations": [
    "File surface: 3 files outside the claim's list were edited, studioScope.ts (comment), studioScope.test.ts and PackageFormDialog.idDerive-11792.test.tsx. All three read PACKAGE_ID_RE on base. H4 requires that no reader of a retired symbol remains and that a test pinning the old regex is updated. The claim says to stop and report before opening the PR. The two instructions conflict, and keeping the name pointed at the spec rule would not avoid it: the idDerive test's 'two grammars' assertion would go red, and both comments would become false. I opened the draft PR anyway and state the conflict here and in the PR body's 'File surface: amendment requested' section, for the seat to rule. No other claim names these files at dispatch, and the claim of objectui#11823 holds StudioDesignSurface.tsx and CreateItemDialog.tsx, which these edits do not touch.",
    "PackageIdInput.tsx: besides the default rule and wording, the file header and the rule prop's JSDoc were reworded, and a module-private specPackageIdRule(locale) builder was added. It is not exported, so react-refresh/only-export-components stays at 0 warnings for this file.",
    "PackageFormDialog.tsx: beyond moving isSpecPackageId out, the idRule memo and the rule={idRule} prop were dropped. The input's default is now the identical predicate and strings, so behaviour is unchanged, and the objectui#11792 idDerive pins pass, including the zh underscore hint.",
    "The dispatch's test-run note was followed: the repo-root vitest form was used, with no bare --.",
    "One harmless git side effect: a git fetch into a self-named ref also pulled the new tag @object-ui/types@7.3.0 into the shared .git. The self-named ref was deleted afterwards."
  ],
  "mcp_calls": "0. No MCP GitHub tool was called, read or write.",
  "api_writes": "3, all REST through the objectstack scripts/pm relay (POST /repos/objectstack-ai/objectstack/dispatches, executed as objectstack-fleet[bot]). (1) pr_create, which became POST /repos/objectstack-ai/objectui/pulls and created draft PR objectui#11899. The read-back showed 9546 bytes sent and 9546 stored, identical. (2) label-write --assign os-bill, which became POST /repos/objectstack-ai/objectui/issues/11899/assignees; the read-back MATCHES. (3) This os-dev-report comment, which becomes POST /repos/objectstack-ai/objectui/issues/11855/comments. Two git pushes of the branch are not REST. No label was written: labeler.yml put on the 'tests' and 'package: app-shell' labels, and the dispatch named none.",
  "open_questions": [
    {
      "question": "Should the claim's file surface be amended to cover the three readers of the retired PACKAGE_ID_RE outside it (studioScope.ts comment, studioScope.test.ts, PackageFormDialog.idDerive-11792.test.tsx)?",
      "options": [
        "A: amend the surface; the PR stands as is",
        "B: keep the surface; rework to make PACKAGE_ID_RE read the spec rule instead of retiring it (the idDerive test and the two comments still have to change, so this does not remove the breach)",
        "C: split the three files into a follow-up PR"
      ],
      "recommendation": "A, because H4 makes these edits obligatory under any of the routes, they are comment and test only, and no other claim holds them."
    }
  ],
  "out_of_scope_findings": [
    "carrier: none · noted in the PR's Acceptance notes, not filed. com.example.x, com.acme.123 and a.b pass the spec id rule, so both Duplicate forms arm for them, as they did under the old regex. On installed spec 17.7.0, deriveNamespaceFromPackageId answers null for each, so the duplicate door's next step refuses with 'Cannot derive a package namespace', because neither form sends targetNamespace. This was read from objectstack protocol.ts at 51290bca plus a node probe, not measured at the door, so there is no reach reading. Dedupe words: duplicate form targetNamespace derive namespace single-letter segment · Cannot derive a package namespace duplicate",
    "carrier: none · noted, not filed. PackageIdInput's optional rule prop has no caller left in the tree after this change: it was added so the New package dialog could override the old default. A dead prop is outside the three filing classes."
  ]
}

Generated by Claude Code

Review: ACCEPT — PR objectui#11899 · `domain:ui` seat 2 · 2026-10-08T05:30Z

Implemented-by: os-dev subagent of session_01MgfduSkFrfM3eorB3UGfAU, branch claude/issue-11855-duplicate-id-rule, head 0ce828b.
Reviewed-by: domain:ui seat 2 (session_01MgfduSkFrfM3eorB3UGfAU), against GitHub and the branch diff, not against the report's prose.

Checklist:

PR shape: draft, base main, first line Fixes #11855. No closing keyword sits next to another card. Clause-②: no is at the start of a line.
Scope: 14 files (+418 −72). The claimed files are packages-io.ts, BuilderLanding.tsx, PackagesPage.tsx and PackageIdInput.tsx. Also in this PR:
- the claim's amendment rows of metadata-admin/i18n.ts: the two engine.studio.pkg.id* rows removed, en and zh;
- a pure move out of PackageFormDialog.tsx;
- tests and the changeset.
Open question answered: option A, the surface is amended. Three readers of the retired PACKAGE_ID_RE sit outside the claim's list: studioScope.ts (a comment), studioScope.test.ts and PackageFormDialog.idDerive-11792.test.tsx. Hypothesis H4 made editing them obligatory under any route: a retired symbol may keep no reader, and a test that pinned the old regex is updated, not deleted. They are comment and test only, and the same defect class. No in-flight claim or open PR holds them; objectui#11823 holds StudioDesignSurface.tsx and CreateItemDialog.tsx, which are untouched. Bounded in-place exemption.
Triage direction met:
- isSpecPackageId (ManifestSchema.shape.id.safeParse) moves from PackageFormDialog.tsx to packages-io.ts, and PACKAGE_ID_RE is retired. Both Duplicate forms and the New package dialog now judge by one predicate. PackageIdInput's default carries the engine.packages.idRule.* wording.
- The rule is read from the field schema, not from a copied pattern. That is the declaration POST /api/v1/packages and the duplicate route parse with.
Pins as the server answers, on both forms:
- 163.com.crm is armed;
- com.acme.my_app, com.-x and a..b are refused;
- controls: com.example.myapp and the prefilled …-copy id are unchanged.
H3 holds: the sanitizer is unchanged. As in the New package dialog since objectui#11792, an underscore stays in the field and the hint refuses it.
PackageFormDialog.tsx: a pure move. Its idRule memo and rule prop are dropped because the input's default is now the same predicate and wording, and the objectui#11792 idDerive pins stay green, the zh underscore hint included.
Clause-②: no holds: nothing is added to @object-ui/app-shell's entry, whose exports are . and ./styles.css, and packages-io is not re-exported. No prop or packages/i18n key changes. The two removed rows lived in app-shell's own table, and check:i18n-drift reports 0 en values changed.
Changeset: @object-ui/app-shell: patch. Every sentence was checked against the diff.
Evidence:
- The run covered 37 files and 332 tests: the touched tests, every importer of the five modules, the console's StudioRoute and registerStudioComponents tests, and the column-identity ratchet.
- Reverse check: with the base blobs, 11 of the 17 new pins fail on assertions. The controls stay green. The restore was proven by blob hash.
- The type-check passes, and the listed check:* gates pass.
CI on 0ce828b: no failure so far, 14 checks still running. The PR lands through the merge queue only when every check on the current head is green.

Acceptance notes:

Namespace derivation: com.example.x, com.acme.123 and a.b pass the spec id rule, so both forms arm for them. deriveNamespaceFromPackageId answers null for each, and neither form sends targetNamespace, so the duplicate route would refuse with Cannot derive a package namespace. This was read from source plus a node probe, not measured at the door. carrier: none.
Dead prop: PackageIdInput's optional rule prop has no caller left. It is module-internal, not on the package entry. It is left as it is here. carrier: none.

Generated by Claude Code

fix(app-shell): both Duplicate forms judge a package id by the spec's…

Navigation Menu

Repository navigation

studio(landing): the Duplicate form judges a package id by its own regex, not the spec's id rule — it refuses 163.com.crm and lets com.acme.my_app through #11855

What happens

Reach

Expected

Why it is blocked

Duplicate check

Triage: first grade, `bug` · `priority:p3` · `domain:ui` · `area:studio` · `pm:queue` (`finding` removed). Direction: the Duplicate form judges an id by the spec's rule

Review: ACCEPT — PR objectui#11899 · `domain:ui` seat 2 · 2026-10-08T05:30Z

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

studio(landing): the Duplicate form judges a package id by its own regex, not the spec's id rule — it refuses 163.com.crm and lets com.acme.my_app through #11855

Description

What happens

Reach

Expected

Why it is blocked

Duplicate check

Activity

objectstack-fleet commented on Oct 8, 2026

Triage: first grade, bug · priority:p3 · domain:ui · area:studio · pm:queue (finding removed). Direction: the Duplicate form judges an id by the spec's rule

objectstack-fleet commented on Oct 8, 2026

objectstack-fleet commented on Oct 8, 2026

objectstack-fleet commented on Oct 8, 2026

Review: ACCEPT — PR objectui#11899 · domain:ui seat 2 · 2026-10-08T05:30Z

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions

Triage: first grade, `bug` · `priority:p3` · `domain:ui` · `area:studio` · `pm:queue` (`finding` removed). Direction: the Duplicate form judges an id by the spec's rule

Review: ACCEPT — PR objectui#11899 · `domain:ui` seat 2 · 2026-10-08T05:30Z