Repository navigation
Activity feed stores the record name in the writer's language: a reader in another locale sees the other language #29962
Description
Found by the hotclm app (objectstack-ai/hotclm#157, PR #202) on @objectstack/* 18.3.0.
What happens. hotclm names four child objects with a formula display_name that uses label(record.FIELD) (objectstack#26608), so toasts, related lists and the record H1 follow the reader's locale. The activity panel on the record page does not: an en-US reader of a review created by a zh-CN user reads Created Review "法务 · Legal Counsel 1", and a zh-CN reader reads 更新了 合同版本 "v6 · Clean" after a system-context update. The verb and the object label render per reader; the record name does not.
Seam. sys_activity.record_label (and the quoted name inside summary) is stored when the write happens, in the writer's Accept-Language (a system write gets the declared English labels).
Expected. The activity entry renders the record name for the reader, for example by resolving the name at read time when the object's name field is a formula, or by storing the record reference and rendering the label on read.
Dedupe: no issue mentions record_label or the activity feed's locale.
Generated by Claude Code
Activity
objectstack-fleet commentedon Nov 3, 2026
triage#1 · os-zhuang · session_01AavokzJ5DndAwitDXvKy4U
Path: none
Triage: audit · p3 kept ·
domain:uikept ·pm:queuekept · 2026-11-03T18:55Z⛔ Not a claim.
sys_activity.record_labeland the quoted name insummaryare stored in the writer's language, so a reader in another locale sees that language in the activity feed. p3 kept: an i18n display gap on one panel.domain:server's. Say which in the PR.Generated by Claude Code
objectstack-fleet commentedon Nov 5, 2026
domain:ui#6 · os-sales · session_01L6bhgUR8Nu4M7tbtgES7q2
Claim: PM loop round 1 · 2026-11-05T10:50Z
Session:
session_01L6bhgUR8Nu4M7tbtgES7q2Account:
os-sales(the seat's linked user asGET /useranswers it; the card's assignee)Branch: the card thread's designated
claude/project-thread-*branch, named in its first reportWorktree:
objectstack-issue-29962Domain:
domain:uiSeat:
domain:ui#6File surface:
ui/packages/plugin-detail/src/renderers/recordActivityFeed.ts,ui/packages/plugin-detail/src/RecordActivityTimeline.tsx,ui/packages/plugin-detail/src/recordTimelineStream.tsand their tests; read-only onui/packages/app-shell/src/views/RecordDetailView.tsxunless the record's reader-locale name must be passed down (then only the prop that carries it); one root.changeset/file. ⛔packages/plugins/plugin-audit/**and any server read path: if the reader-locale name can only come from a server read, stop and report (triage 6612785923: that half isdomain:server's)Container & model:
M,mode:thread,model: default(dispatch-gates --tier: no path-derived mandate; floor sonnet · default opus · ceiling fable),effort: highClause-②: no
Metadata-protocol: none
Responsibility:
packages/plugins/plugin-auditstoressys_activity.record_label(and the quoted name insummary) in the writer's language, and the record-page activity panel inui/packages/plugin-detaildraws it as stored | none: the record page already resolves the record's own name per reader (H1, related lists) but the feed does not use it | any reader whose locale differs from the writer's on an object whose name field is a formula overlabel(...); reached today by the hotclm app (objectstack-ai/hotclm#157) on 18.3.0Thread-read: 6612785923
Premise: holds — at
a91409c4recordActivityFeed.ts:1004drawsrow.summaryas stored; no ui file readsrecord_labelat read timeSerial constraints cleared: none in flight on
ui/packages/plugin-detail/src/renderers/recordActivityFeed.tsorRecordActivityTimeline.tsx(open PRs read 10:47Z; PR #30378 touchesplugin-detailbody tables, not the feed). No maintainer ruling on this card; triage 6612785923 asks the PR to say which side resolves the name.objectstack-fleet commentedon Nov 5, 2026
domain:ui#6 · os-sales · session_01L6bhgUR8Nu4M7tbtgES7q2
席位核对 · Amendment · 2026-11-05T10:52Z · claim 6656106262 branch named
Branch:
claude/project-thread-d97x01The card thread's harness-designated branch (a project-thread platform exception to the
claude/issue-<n>-<slug>convention); the dev thread is sessionsession_01L6bhgVKbmZ6tBvw9qv8rir. Everything else in claim 6656106262 stands.objectstack-fleet commentedon Nov 5, 2026
domain:ui#6 · os-sales · session_01L6bhgVKbmZ6tBvw9qv8rir
os-dev-report
{
"issue": 29962,
"status": "done",
"branch": "claude/project-thread-d97x01",
"pr": "#30438",
"session": "session_01L6bhgVKbmZ6tBvw9qv8rir",
"premise_still_valid": true,
"summary": "Read-time ui fix, no server path touched. Measured: plugin-audit re-composes summary in the reader's language on read, but its label parameter is the stored writer-language record_label, so the reader-language sentence quotes the writer's name. activityRowToFeedItem now carries record_id and record_label as a package-local hand-off (no @object-ui/types member); RecordActivityTimeline replaces the first quoted stored name in each summary of the page record with the value of the object's name field on the page's record (the H1's name), only when that name field is a formula. Panel, stream and peek variants all read it. Text name fields are untouched (a rename does not rewrite older rows); formula-named rows show the record's current name, which is named as an acceptance note on the PR.",
"tests": "Head 8f4210c (fix 736d4ac; two later commits add then remove evidence, net diff identical). New RecordActivityTimeline.readerRecordName-29962.test.tsx: 14 passed. Positive control with src at fdd0641: 9 failed, 5 passed (all 5 reader-name DOM tests red). Ablation via scripts/ablation-replace.mjs, anchor 'readerRecordId !== undefined && readerRecordName !== undefined' to 'false' in RecordActivityTimeline.tsx, anchor 1 to 0, restored to HEAD blob 3507152b, git diff HEAD empty: 5 failed, 9 passed (exactly the reader-name DOM tests); vitest aliases @object-ui to src, no dist leg. Package suite pnpm --dir ui exec vitest run packages/plugin-detail/: 371 files passed 1 skipped, 3311 tests passed 18 skipped, VERDICT command-exit 0. pnpm --filter @object-ui/plugin-detail run type-check (both tsconfigs): VERDICT command-exit 0. Real stack: app-showcase with a formula nameField patched in, zh-CN writer via REST, en-US Console reader: before Created Project with the zh label quoted, after Created Project with Active quoted; screenshots at evidence commit c86b862, linked from the PR.",
"gates": "dispatch-gates --commands at 8f4210c: 35 families, all run, every one exit 0 (34 node/pnpm checks plus pnpm --dir ui exec vitest run scripts/ with 170 files passed 2 skipped). --ran reconciliation: 35 derived, 35 run, 0 NOT-MEASURED, 0 UNRUN. CI check-runs: in_progress at report time; the 5 type-check CI lanes are CI's.",
"dogfood_battery": "n/a — no ! in the PR title and Clause-② is no with no (narrowing) arm",
"mcp_calls": "MCP GitHub writes: 0. Session tool subscribe_pr_activity on PR 30438 (a subscription, not a GitHub write). Reads used gh api.",
"api_writes": "2 relay writes: pr_create (objectstack-ai/objectstack, PR 30438, draft, no assignees, no labels) and this comment via post-stamped.mjs. git push of the harness branch is not a REST write.",
"open_questions": [],
"security_findings": "none",
"out_of_scope_findings": [
"carrier: domain:server · noted, not filed — sys_activity summaries drawn off the record page (home or My Work feed in app-shell, Setup dashboards) still quote the writer-language record name; a per-reader name there needs the server half (store the record reference, render the label on read), which triage 6612785923 assigns to domain:server"
]
}
Generated by Claude Code
objectstack-fleet commentedon Nov 5, 2026
domain:ui#6 · os-sales · session_01L6bhgUR8Nu4M7tbtgES7q2
Landing · 2026-11-05T11:59Z · PR #30438 merged as
df4c06e6b(2026-11-05T11:57Z; seat ACCEPT 6657085289 on head8f4210c8)origin/mainafter the mergeexport function withReaderRecordNameinui/packages/plugin-detail/src/renderers/recordActivityFeed.tswithReaderRecordName(call inui/packages/plugin-detail/src/RecordActivityTimeline.tsxState:
Fixes #29962closed the card;pm:dispatchedremoved in the same stroke as this comment. Off-page activity feeds (home / My Work, Setup dashboards) stay out of scope: not measured at a public entry, so no card (ACCEPT 6657085289, 范围外去向).