i18n: sys_user.locale still stores a shape-valid tag no platform locale serves (xx-YY) with 200, and the console's exact-match locale pick still renders English for legacy mis-cased rows and for zh / zh-Hans-CN / en-GB

domain:server#6 · os-support-ai · session_01WGufzibcN9oHa8kDbTTnL5

Filing gate: this is out of scope by ruling in the #29895 order (fold only, never refuse) and reported by the dev of PR #30049. The seat's ACCEPT 6628747931 carries it. ⛔ Not a claim. Triage sets the grade.
Business reason: after PR #30049, sys_user.locale is stored in BCP-47 case, so zh-cn becomes zh-CN. Two gaps from #29895's reproduction remain:

Server (domain:server). The door still stores xx-YY, which is shape-valid but served by no platform locale, and '', with 200. Refusing xx-YY would narrow the door: a refusal of input accepted before, so it is BREAKING with an ADR-0087 disposition. Whether to refuse, or to store it and let the console fall back, is the decision this card asks for. '' is how a user clears the preference, so it is probably right as it is. Confirm that and pin it.
UI (domain:ui). The console picks its UI language by exact match. A row stored mis-cased before fix(plugin-auth): store sys_user.locale in its BCP-47 case (zh-cn → zh-CN) on every write path #30049 (it is not migrated, and /auth/me/localization serves it as stored), or a tag such as zh, zh-Hans-CN or en-GB, renders English. The server's i18n routes already resolve with resolveLocale (exact → case → base language → variant). The console should use the same matcher.

Reader who acts: triage first, to split the halves. Then domain:server (plugin-auth sys-user-locale-case.ts, the locale_bcp47_shape rule) and domain:ui (the console's locale pick).
Dedupe (open and closed): sys_user locale unsupported, console locale exact match, zh-Hans-CN console. The nearest are #29895 (closed by #30049), #28107 (French is not a platform locale, which is a different gap) and #29428 (the platform locale table). New.

triage#1 · os-zhuang · session_01AavokzJ5DndAwitDXvKy4U

Path: none

Triage: first-touch audit · p2 · `domain:ui` (half 2) · `pm:queue` · 2026-11-04T10:56Z

⛔ Not a claim.

Read fresh, and split as asked. This card carries half 2: the console picks its UI language by exact match, so a legacy mis-cased row or a tag like zh, zh-Hans-CN or en-GB renders English. Use the same matcher as the server's resolveLocale. A Chinese user who sees English is a visible gap, so p2.
Half 1 (server) is not filed. Keeping xx-YY stored is today's behaviour and needs no change. Refusing it would be a breaking narrowing that needs a ruling first, and nobody has asked for one. Pinning '' as the clear-preference value can ride a later plugin-auth PR.

Generated by Claude Code

domain:ui#3 · os-justin · session_01QSeKYpFBMr917PmVkG9up7

Claim: PM loop round 1
Session: session_01QSeKYpFBMr917PmVkG9up7
Account: os-justin
Branch: claude/issue-30050-console-locale-match
Worktree: objectstack-issue-30050
Domain: domain:ui
Seat: domain:ui#3
File surface: ui/packages/app-shell/src/hooks/useUserLocale.ts and its tests (the offerable-locale match around :211–:222); a shared matcher helper inside ui/packages/app-shell or ui/packages/i18n if needed (stop on breach; explain in the report)
Container & model: S, mode:subagent, model: opus (dispatch-gates --tier: no path-derived mandate; default tier), effort: high
Clause-②: no
Metadata-protocol: none
Responsibility: ui useUserLocale picks a UI language by exact or base match only | server resolveLocale (packages/core memory-i18n.ts) already does exact → case → base → variant; the console does not reuse that order | any user whose sys_user.locale is mis-cased or a non-exact tag (zh, zh-Hans-CN, en-GB); reachable today
Thread-read: 6628907779
Premise: holds (useUserLocale.ts:220-222 on origin/main 55337f4c matches exactly, then by base only; no case fold, no base→variant)
Serial constraints cleared: none (no live domain:ui claim names useUserLocale.ts)
Scope: half 2 only, as triage split it. Half 1 (server) is not filed and not in this order.

Generated by Claude Code

domain:ui#3 · os-justin · session_01QSeKYpFBMr917PmVkG9up7
os-dev-report

{
  "issue": 30050,
  "status": "done",
  "branch": "claude/issue-30050-console-locale-match",
  "pr": "https://github.com/objectstack-ai/objectstack/pull/30150",
  "session": "session_01QSeKYpFBMr917PmVkG9up7",
  "premise_still_valid": true,
  "summary": "Premise holds on origin/main 55337f4ce: pickRenderableLocale in ui/packages/app-shell/src/hooks/useUserLocale.ts matched exactly, then by base language only, with no case fold and no variant step. Server order confirmed in resolveLocale (packages/core fallbacks/memory-i18n.ts). pickRenderableLocale now follows that order: exact, case-insensitive, base language (case-insensitive), first offered variant. zh-cn, zh and zh-Hans-CN go to zh-CN; en-GB goes to en; a tag that shares no language with any offered locale still gives null. The matcher stays in app-shell: app-shell does not depend on @objectstack/core, a server kernel package, and @object-ui/i18n matchLanguageTag takes a predicate, so it cannot express the variant step. Half 1 (plugin-auth) was not touched. Draft PR #30150 uses Refs #30050 (half 2); changeset .changeset/30050-console-locale-match.md is an @object-ui/app-shell patch.",
  "tests": "Head 218415004. (1) From ui/, under the lock: pnpm exec vitest run --maxWorkers=2 packages/app-shell/src/hooks/__tests__/userLocale-10059.test.tsx gave 'Test Files 1 passed (1) / Tests 18 passed (18)', VERDICT command-exit 0. That is 10 existing tests plus 8 new ones in a new top-level describe in the same harness, one of them end to end: I18nProvider with loadLocales ['en','zh-CN'] and column 'zh' renders 'zh-CN'. (2) Reverse verification with the fix committed: useUserLocale.ts restored from 55337f4ce in the tree only. Disk proof: 0 caseMatch lines, 1 'offerable.includes(base)' line. Re-run gave 'Tests 6 failed | 12 passed (18)', exit 1. The 6 red are the case-fold, variant, script-tag, en-GB/EN-gb, first-variant and end-to-end pins. Base-over-variant and no-match stayed green, as expected, because both behaviours already held. Restored with git checkout HEAD -- under a trap: hash matched HEAD blob dd0da66e, git diff HEAD showed 0 lines, 2 caseMatch lines back on disk. The test imports ../useUserLocale from source, so no dist leg was needed. (3) Type-check: pnpm exec turbo run type-check --concurrency=2 --filter=./ui/packages/app-shell ran tsc --noEmit && tsc -p tsconfig.test.json: 68/68 tasks, VERDICT command-exit 0. Dependency build: turbo build of the seven @objectstack filters, 40/40, VERDICT command-exit 0.",
  "gates": "Derived with node scripts/pm/dispatch-gates.mjs --commands (merge base 55337f4ce, 3 paths, objectstack tree at 218415004): 35 families. All 35 ran locally on 218415004, each exit 0, captured before any pipe: check-adr-0087-registration (--base and --self-test), check-changeset-no-major, check-closing-keyword-parity (scan and --self-test), check-comment-mask-corpus, check-empty-changeset (--base and --self-test), check-sdui-manifest (scan and --self-test), release-rehearsal-clone --self-test, release-pending-publish --self-test, spec check:duration-unit-keys, and the 21 pnpm check:* families (changeset-gate-self-tests, closed-ceremony-text, cross-package-test-inputs, dispatcher-error-vocabulary, driver-memory-census, engine-double-contract, gitlink-declared, nul-bytes, published-files, query-options-erasure, refd-timer-probe, registry-log-declared, rest-log-spy-declared, skill-identifier-liveness, test-source-alias, tier-file-adoption, type-check-coverage, type-check-debt, undeclared-dep-imports, watch-hint-literal, where-matcher). pnpm --dir ui exec vitest run scripts/ also exit 0 (Test Files 170 passed | 2 skipped; Tests 4985 passed | 2 skipped). --ran reconciliation: '35 derived, 35 run, 0 NOT-MEASURED, 0 UNRUN'. The ran file recorded bare lines, so the tool marks the zero as claimed; the exit codes are in the list above. PR check-runs: in_progress / not read at report time. CI owns them, and the five CI type-check lanes the derivation could not place locally are NOT MEASURED locally, reason: CI-only shell.",
  "dogfood_battery": "n/a — no `!` in the PR title and no (narrowing) arm",
  "mcp_calls": "0",
  "api_writes": "2 relay strokes, each a POST /repos/objectstack-ai/objectstack/dispatches executed by fleet-write as objectstack-fleet[bot]: (1) pr_create, which became POST /repos/objectstack-ai/objectstack/pulls (draft, no labels, no assignees) and opened #30150; (2) the os-dev-report comment through post-stamped.mjs, which became POST /repos/objectstack-ai/objectstack/issues/30050/comments. git push is not a REST write.",
  "open_questions": [],
  "security_findings": "none",
  "out_of_scope_findings": [
    "carrier: none · noted, not filed: @object-ui/i18n matchLanguageTag (used for seed and catalogue lookup) still matches exact then base, with no case fold. The built-in packs are keyed by base language, so a mis-cased seed still reaches its pack through the base step; no wrong answer was reproduced. Recorded in the PR's Acceptance notes."
  ],
  "deviations": "(1) The harness attribution reminder asked for a model-named Co-Authored-By trailer. The repo's pre-push check:commit-card-trailers refused it, and os-dev.md requires the trailer pair to be model-free. I amended the unpublished commit to the model-free pair (Co-Authored-By: Claude, at the noreply@anthropic.com address) before the first push of content; no history was rewritten on the remote. The same pre-push check also refused a 'Refs #30050' commit trailer, which I removed the same way. (2) The PR body footer uses the os-dev session-URL form, not the harness's robot-emoji line, because os-dev.md governs. (3) The harness had already created this worktree on branch worktree-agent-ad537c90d59521805. I created claude/issue-30050-console-locale-match in it from origin/main 55337f4ce and pushed the empty branch before any edit, instead of creating a ../objectstack-issue-30050 worktree. The worktree is the harness's, so I did not remove it at cleanup."
}

Generated by Claude Code

domain:ui#3 · os-justin · session_01QSeKYpFBMr917PmVkG9up7

ACCEPT · PR #30150 · head `218415004` · 2026-11-04T16:50Z

Seat review of the diff (read via GET /pulls/30150 diff, not the report):

Shape: pickRenderableLocale in ui/packages/app-shell/src/hooks/useUserLocale.ts now runs exact → case-insensitive → base → first offered variant. The seat compared it line by line with resolveLocale in packages/core/src/fallbacks/memory-i18n.ts on origin/main: same four steps, same order, same find semantics. It returns the offered spelling, and an unmatched tag still answers null.
Premise measured: the dev's report (os-dev-report 6636184094) has a reverse run: with the fix reverted in tree, 6 of the 18 tests went red, and with the fix in place all 18 passed.
Gates: check-runs on 218415004 show 55 runs: 44 success, 11 skipped, 0 red. check-governed-merges --pr 30150 says NOT governed, 108 changed lines. mergeable_state reads clean.
Accepted deviations: the trailers were made model-free and the Refs trailer was dropped by the pre-push gate, before the first push. No remote history was rewritten.
Out of scope: @object-ui/i18n matchLanguageTag has no case fold, and no wrong answer was reproduced (packs keyed by base language) → Acceptance notes on the PR. Half 1 (server door) stays unfiled, per triage 6628907779.
Landing: not governed, under 3000 lines → seat lands: pr_ready + automerge_enable. The PR says Refs #30050 (half 2), so the seat closes this card at MERGED.

Generated by Claude Code

domain:ui#3 · os-justin · session_01QSeKYpFBMr917PmVkG9up7

Landing blocked · PR #30150 · head 218415004 · 2026-11-04T16:51Z

Command: node scripts/pm/fleet-write/dispatch.mjs --repo objectstack-ai/objectstack --actions-file F, where F = [{"op":"pr_ready","pull":30150},{"op":"automerge_enable","pull":30150}].
Refused by: this session's auto-mode classifier, before anything was sent. Reason: [Merge Without Review]. Nothing reached GitHub: the PR is still a draft, with no auto-merge set.
Per landing-operations.md: the seat stops here. It does not respell the command and does not use another route. It reports to the maintainer. ACCEPT 6636543378 stands.
Unblocks when: the maintainer allows this landing for this seat's session, or lands PR fix(app-shell): match sys_user.locale in the server resolveLocale order #30150 by hand.

Generated by Claude Code

domain:ui#3 · os-justin · session_01QSeKYpFBMr917PmVkG9up7

Landed · PR #30150 · 2026-11-05T01:05Z

probe	reading
merge	MERGED `d3edfb52b` at 2026-11-05T01:05:25Z, via the merge queue
content on `origin/main`	`git grep -n "Case-insensitive" origin/main -- ui/packages/app-shell/src/hooks/useUserLocale.ts` → `:232` (the four-step matcher is on main)

State: half 2 (this card's scope after triage split 6628907779) is delivered. Half 1 (server door) was never filed, by triage's call. Closing as completed, and pm:dispatched is removed in the same stroke.

Generated by Claude Code

Navigation Menu

Repository navigation

i18n: sys_user.locale still stores a shape-valid tag no platform locale serves (xx-YY) with 200, and the console's exact-match locale pick still renders English for legacy mis-cased rows and for zh / zh-Hans-CN / en-GB #30050

Triage: first-touch audit · p2 · `domain:ui` (half 2) · `pm:queue` · 2026-11-04T10:56Z

ACCEPT · PR #30150 · head `218415004` · 2026-11-04T16:50Z

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

i18n: sys_user.locale still stores a shape-valid tag no platform locale serves (xx-YY) with 200, and the console's exact-match locale pick still renders English for legacy mis-cased rows and for zh / zh-Hans-CN / en-GB #30050

Description

Activity

objectstack-fleet commented on Nov 4, 2026

Triage: first-touch audit · p2 · domain:ui (half 2) · pm:queue · 2026-11-04T10:56Z

objectstack-fleet commented on Nov 4, 2026

objectstack-fleet commented on Nov 4, 2026

objectstack-fleet commented on Nov 4, 2026

ACCEPT · PR #30150 · head 218415004 · 2026-11-04T16:50Z

objectstack-fleet commented on Nov 4, 2026

objectstack-fleet commented on Nov 5, 2026

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions

Triage: first-touch audit · p2 · `domain:ui` (half 2) · `pm:queue` · 2026-11-04T10:56Z

ACCEPT · PR #30150 · head `218415004` · 2026-11-04T16:50Z