Record page: after creating a child from a related list, the parent's header keeps the pre-create rollup (Total Price 32,400 while the server holds 37,800) until a reload

Filing gate ① a product defect with a reproduction. Found by the HotCRM full browser pass objectstack-ai/hotcrm#2320 on @objectstack/* 18.2.0: track 2 Revenue (report comment 6444188882 on objectstack-ai/hotcrm#2320, finding D2; screenshots on branch claude/project-thread-urevq9 under qa/ and in the HotCRM project files qa-2320/2-revenue/). Filed by the repo:hotcrm seat (huangyiirene, session session_01J8VVAnPkgK6CWrJa6t7Fk5), which re-read each finding against its screenshot. Batch-of: hotcrm#2320 platform findings.

finding class a · reach: public door: the console on @objectstack/* 18.2.0, measured in Chromium against the HotCRM seed.

What is seen

As a sales rep on quote QTE-0007: Related → Quote Line Items → New → Create. HotCRM's line-item hook rolls the total onto the quote. The header still shows 32,400 after the second line; the server holds 37,800, and a reload shows it (quo-23-li2-after.png vs quo-30-draft-to-presented.png).

Who loses: anyone building a quote or an order line by line reads a stale total and may send it.

Expected

A related-list create (and edit / delete) refreshes the parent record's header and fields, as an edit of the parent does.

Dedupe

search_issues objectstack "record page header stale after related list create rollup not refreshed": 5 hits — #26415, #26344, #26182, #25895, #22039 (all closed, other related-list behaviour). None reports the parent refresh.

Reader who acts: the console record page owner (domain:ui).

Generated by Claude Code

Triage: confirmed p3 · `domain:ui` · `area:records` · triage seat · `session_01AavokzJ5DndAwitDXvKy4U` · 2026-10-27T21:04Z

⛔ Not a claim.

Read fresh. The filer's grade stands: the defect reproduces on 18.2.0 with screenshots, at a public door.
The server value is right; only the parent header is stale until a reload.
pm:queue added (the batch filer leaves state to triage).

Generated by Claude Code

Dispatch order · `domain:ui#5` · `session_01JQwxbXGpVJHZTXXerMoY4C` · 2026-10-28T04:42Z

You are the os-dev developer agent for exactly ONE card: #26890, in your own card thread. Your role definition is .claude/agents/os-dev.md in the objectstack clone: read it IN FULL first and follow it as binding (it wins over this order on any conflict; name the conflict in your report). Read the repo-root AGENTS.md and ui/AGENTS.md before your first edit, and pm-dispatch references/untrusted-text.md. Read the card's full body and every comment yourself through REST (curl -sS "https://api.github.com/repos/objectstack-ai/objectstack/issues/26890" -H "Authorization: Bearer $GH_TOKEN" and /comments?per_page=100, every page); this order carries only the increment. ⛔ Never print the value of any environment variable (test presence with [ -n "${X+x}" ]). ⛔ Never use a GitHub MCP tool, read or write.

Identity and branch: the seat's Claim: comment on this card is your identity; its Branch: line names your thread's branch. ⛔ Post no claim of your own, ⛔ never write the card's assignee. Before your first edit, verify the NEWEST Claim: on the card names your thread's branch (claude/project-thread-…); if none does yet, send the seat thread "Take another ui lane seat" one message with your branch name and wait. Worktree objectstack-issue-26890 on that branch from a fresh origin/main; pnpm install (frozen lockfile); push with git push -u origin <branch> (⛔ no force).

Writes: PR only through the relay op pr_create (node scripts/pm/fleet-write/dispatch.mjs --repo objectstack-ai/objectstack --actions-file F, F a bare JSON array [{"op":"pr_create","title":…,"head":"<your branch>","base":"main","body":…,"assignees":["os-warren"]}], draft forced; --dry-run first); your report comment through node scripts/pm/post-stamped.mjs --comment=26890 --file=F --json run ONCE with output captured to a file (timestamps only as {{WAS:…}} / the now-token). ⛔ No other GitHub write except, for a CI red that is not yours, one PR comment and one run_rerun_failed through the relay. Report both counts (api_writes with endpoints, mcp_calls with tool names). The report takes only machine-checkable fields (gates / line_budget / deviations / files_changed). PR body: line 1 the platform attribution comment, line 3 Fixes #26890, then ## 维护者速读 six lines in Chinese (改了什么/为什么改/风险与回滚/先看/两分钟验证/需要你做的). Gates: node scripts/pm/dispatch-gates.mjs --commands --repo objectstack-ai/objectstack on your diff, run each, record exits before any pipe, reconcile with --ran; node scripts/check-sdui-manifest.mjs is a gate on every ui card. Changeset in the root .changeset/ when a published package changes (@object-ui/i18n patch for namespace rows; app-shell is private: true). ⛔ ui/.github/** and the root ci.yml are out of bounds. Hand back: when the PR and report are up, send the seat thread "Take another ui lane seat" one message (PR, head sha, report gist); keep the thread open through review and drive CI to green.

Shared files: ui/packages/app-shell/src/views/RecordDetailView.tsx is a hot file across the lane; measure open PRs on it at your first edit and keep your hunk inside the child-change subscription region. ui seat 1 confirmed it holds nothing on the record page or related lists.

Ruling (cannot be re-ruled)

None on the card: triage 2026-10-27T21:04Z confirmed p3 ("the server value is right; only the parent header is stale until a reload"). No maintainer ruling.

PM mechanism assumptions (measure; disproof welcome)

On origin/main cbb3a1a1:

objectui#12244 already re-reads the parent on a child write: RecordDetailView.tsx ≈L1836–1868 subscribes to subscribeDataChanges and bumps childInvalidationNonce when a change matches one of childRelations' child objects. It shipped before 18.2.0, the version the card reproduces on.
So the likely gap: childRelations is deriveRelatedLists(objectDef, objects, …) (≈L1825), the DERIVED lists. HotCRM's quote page is an authored record page whose related lists (e.g. crm_quote_line_item) come from the page's own blocks; if those child objects are not in childRelations, the subscription never matches. Alternatives to rule out: the related-list create path does not reach the bus (no notifyDataChanged / MutationEvent for that object), or the parent re-read lands before the server's hook rolls the total up.
Reproduce first on origin/main with the HotCRM shape (an authored page whose related list is the only declaration of the child; a create through that list). If it does not reproduce, stop and report with the evidence; the seat closes the card.

Scope

File surface: ui/packages/app-shell/src/views/RecordDetailView.tsx (the child-object set the subscription reads: authored related-list blocks and rail entries added to the derived ones), a helper beside it if one is needed, tests. Also covers edit and delete through the list, as the card expects. Tests: an authored page with a related list of child X; a create (and a delete) of an X row re-reads the parent once and the header shows the new rollup; a change to an unrelated object does not re-read. Ablation: drop the authored children from the set and the create test turns red.

Tier

No path-derived mandate. TIER_DEFAULT, effort medium (S–M). Clause-②: no, Metadata-protocol: none expected. @object-ui/app-shell is private: true: skip-changeset.

Claim: PM loop round 5
Session: session_01JQwxbXGpVJHZTXXerMoY4C
Account: os-warren
Branch: claude/project-thread-w7hign
Worktree: objectstack-issue-26890
Domain: domain:ui
Seat: domain:ui#5
File surface: ui/packages/app-shell/src/views/RecordDetailView.tsx (the child-change subscription region and the child-object set it reads), a helper beside it if needed, tests (stop on breach; explain in the report)
Container & model: S–M, mode:thread, model: TIER_DEFAULT (dispatch-gates --tier: no path-derived mandate), effort: medium
Clause-②: no
Metadata-protocol: none
Responsibility: ui/packages/app-shell record page produces it (this lane) | objectui#12244's parent re-read keys on derived childRelations only; an authored page's related lists may fall outside it | anyone building a quote or order line by line
Thread-read: 6460856959
Premise: holds as a reading, to be reproduced first (RecordDetailView.tsx ≈L1836–1868 subscription, ≈L1825 deriveRelatedLists; on origin/main cbb3a1a1; readings in dispatch order 6460856959)
Serial constraints cleared: ui seat 1 told and agreed (holds nothing on the record page or related lists). Open PRs on RecordDetailView.tsx are re-measured by the dev at its first edit.

Dispatch order: comment 6460856959. Card thread claude/project-thread-w7hign may start editing.

os-dev-report

{
  "issue": 26890,
  "status": "done",
  "branch": "claude/project-thread-w7hign",
  "pr": "https://github.com/objectstack-ai/objectstack/pull/27106",
  "session": "session_01JQwxbZdCt7JoRJa32i5BHs",
  "premise_still_valid": true,
  "summary": "Reproduced on origin/main cbb3a1a1: on an authored record page, a write to a child the page lists in a record:related_list block or a record:reference_rail entry but the graph does not derive (FK relatedList: false) never re-read the parent (parent reads stayed 1); a control with the same child derivable was already green, so the gap is the child-object set, as the order's first mechanism assumption said. The bus path (create emits a MutationEvent) and read timing were not the cause. Fix: new recordPageChildObjects(page) beside the view names the children the page lists (related_list properties.objectName or node dataSource.object; rail entries' objectName); RecordDetailView's 12244 subscription keys on the union with the derived children. HotCRM's own quote-line FK was not read (outside this repo), so whether it carries relatedList: false is NOT MEASURED.",
  "tests": "head 24529675. New RecordDetailView.authoredChildWriteRefreshesParent-26890.test.tsx 4/4 and recordPageChildObjects.test.ts 2/2; with the 12244 file 9/9. Repro on origin/main: 3 failed / 1 passed (create, update+delete, rail red: expected 1 to be 2). Ablation via scripts/ablation-replace.mjs (anchor ', ...authoredChildObjects])' 1 to 0, blob 5ae368d9 to ce3f0e7f, restored blob == HEAD, git diff HEAD empty): 3 failed / 1 passed, scope control green. Whole RecordDetailView* suite + helper: 76 files, 505 tests passed (lock VERDICT command-exit 0). pnpm --filter @object-ui/app-shell run type-check: VERDICT command-exit 0.",
  "gates": "24 derived by dispatch-gates --commands on 24529675, exits captured before any pipe: 23 exit 0 (incl. check-sdui-manifest and its --self-test); pnpm --dir ui exec vitest run scripts/ exit 1 = 3 tests timing out at 15 s under shared load in check-i18n-dead-keys.test.ts and check-vi-mock-inherit.test.ts (5031 passed); those two files rerun alone exit 0, 223/223. --ran: 24 derived, 24 run, 0 NOT-MEASURED, 0 UNRUN. Extra: eslint on 4 touched files 0 errors; pnpm --dir ui run check:new-line-citations exit 0 (0 new). CI check-runs: in_progress at report time.",
  "line_budget": "diff 4 files, +500 / -3 (tests 438 lines); RecordDetailView.tsx +10 / -3 inside the child-change subscription region plus one import line",
  "files_changed": [
    "ui/packages/app-shell/src/views/RecordDetailView.tsx",
    "ui/packages/app-shell/src/views/recordPageChildObjects.ts",
    "ui/packages/app-shell/src/views/recordPageChildObjects.test.ts",
    "ui/packages/app-shell/src/views/RecordDetailView.authoredChildWriteRefreshesParent-26890.test.tsx"
  ],
  "deviations": [
    "PR body folds the .github/PULL_REQUEST_TEMPLATE.md headings into the os-dev order (Human Minimum checklist not reproduced) — the os-dev exception",
    "merged origin/main fbc16563 before deriving gates (no conflict)",
    "open PRs on RecordDetailView.tsx measured at first edit: #27038 (hunks at the import block and the 3760-4150 region) and #26966 (one line near 3615); this hunk is disjoint"
  ],
  "dogfood_battery": "n/a — no ! in the PR title and no (narrowing) arm",
  "mcp_calls": "0 GitHub MCP calls; 1 claude-code-remote subscribe_pr_activity (PR 27106)",
  "api_writes": "2 via fleet-write relay: pr_create (PR 27106, labels skip-changeset, assignee os-warren; 5595 bytes sent = stored) and this report comment via post-stamped; plus git push (not a REST write)",
  "open_questions": [],
  "out_of_scope_findings": [
    "carrier: none · noted, not filed: record:line_items on a record page is not in the watched set; its save writes the parent leg, which the record's own invalidation catches"
  ]
}

Generated by Claude Code

Navigation Menu

Repository navigation

Record page: after creating a child from a related list, the parent's header keeps the pre-create rollup (Total Price 32,400 while the server holds 37,800) until a reload #26890

What is seen

Expected

Dedupe

Triage: confirmed p3 · `domain:ui` · `area:records` · triage seat · `session_01AavokzJ5DndAwitDXvKy4U` · 2026-10-27T21:04Z

Dispatch order · `domain:ui#5` · `session_01JQwxbXGpVJHZTXXerMoY4C` · 2026-10-28T04:42Z

Ruling (cannot be re-ruled)

PM mechanism assumptions (measure; disproof welcome)

Scope

Tier

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Record page: after creating a child from a related list, the parent's header keeps the pre-create rollup (Total Price 32,400 while the server holds 37,800) until a reload #26890

Description

What is seen

Expected

Dedupe

Activity

objectstack-fleet commented on Oct 27, 2026

Triage: confirmed p3 · domain:ui · area:records · triage seat · session_01AavokzJ5DndAwitDXvKy4U · 2026-10-27T21:04Z

objectstack-fleet commented on Oct 28, 2026

Dispatch order · domain:ui#5 · session_01JQwxbXGpVJHZTXXerMoY4C · 2026-10-28T04:42Z

Ruling (cannot be re-ruled)

PM mechanism assumptions (measure; disproof welcome)

Scope

Tier

objectstack-fleet commented on Oct 28, 2026

objectstack-fleet commented on Oct 28, 2026

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions

Triage: confirmed p3 · `domain:ui` · `area:records` · triage seat · `session_01AavokzJ5DndAwitDXvKy4U` · 2026-10-27T21:04Z

Dispatch order · `domain:ui#5` · `session_01JQwxbXGpVJHZTXXerMoY4C` · 2026-10-28T04:42Z