Default list view synthesized from highlightFields omits the nameField — the showcase Account list shows 14 rows with no name column

[hotlong]

What happens

Showcase → Data Model → 客户 (showcase_account), default view 所有记录: the grid columns are #, 生命周期, 行业, 年收入, 操作. There is no name column, so the 14 rows cannot be told apart. The object declares nameField: "name" (name is a required text field, label "Account Name") and highlightFields: ["status", "industry", "annual_revenue"], and it declares no list views (/api/v1/meta/object/showcase_account has no listViews; /api/v1/meta/view has no view for it).

So the console synthesizes the default view from highlightFields alone and never prepends the object's nameField. Objects that declare explicit views (Task) are fine.

Expected

A synthesized default list always starts with the nameField column (as the row's link), followed by highlightFields / the first N visible fields. A list where the row cannot be identified is not a usable default.

Evidence

Screenshot of the grid + the two API responses above; objectui @ 67dadd6, objectstack main @ a39b02a6b.

Found during the objectstack pin-bump dogfood (objectstack#14295).

Generated by Claude Code

[hotlong]

Claim: PM loop round 1 (dogfood dispatch)
Session: session_62849a16-0144-4728-8942-9f60bb3a73f1
Branch: claude/issue-7245-default-view-namefield
Worktree: objectui-issue-7245
Domain: domain:ui
File surface: the default list-view synthesis (expected under packages/app-shell or packages/plugin-list — the code that builds 所有记录 from highlightFields) + its tests, .changeset/ (stop on breach; explain in the report)
Container & model: M, model: opus, mode:subagent
Clause-②: no
Serial constraints cleared: none — open PR #7261 (plugin-list FLS $select) touches ListView fetch fields, not the column synthesis; re-check overlap when both are in flight

Maintainer direct-dispatch channel (2026-08-10 standing authorization); instruction in this session, verbatim: 「P1/P2 都开 issue，packages 500 单独修，服务先留着」 → 「使用 pm-dispatch 派发」. Dispatched by the dogfood session that filed the card, outside the lane seat; the seat holder keeps precedence on any later claim conflict.

Generated by Claude Code

[hotlong]

Claim: os-dev (dispatched, round 1)
Session: session_62849a16-0144-4728-8942-9f60bb3a73f1
Branch: claude/issue-7245-default-view-namefield
Worktree: objectui-issue-7245 (branched from origin/main @ adbda1b)

Starting on the default list-view column synthesis. Will not touch the assignee field.

Generated by Claude Code

[hotlong]

os-dev-report

{
 "issue": 7245,
 "status": "done",
 "branch": "claude/issue-7245-default-view-namefield",
 "pr": "https://github.com/objectstack-ai/objectui/pull/7284",
 "premise_still_valid": true,
 "summary": "Confirmed the premise against the running showcase backend before writing code: served /api/v1/meta/object/showcase_account carries nameField 'name', highlightFields [status, industry, annual_revenue] and NO listViews/list_views/list, and /api/v1/meta/view has 0 of its 35 views bound to the object. One wording correction: the object source (examples/app-showcase/src/data/objects/account.object.ts) does NOT declare nameField — the platform stamps it onto served metadata — so the card's 'the object declares nameField' is imprecise; the served reading the console consumes is exactly as reported, so the diagnosis stands. PM assumption 2 confirmed: THREE synthesis faces read highlightFields verbatim, not one — app-shell ObjectView.tsx defaultListColumnsFromObject (cap 5, the reported surface), app-shell InterfaceListPage.tsx defaultColumnsFromObject (cap 6), and plugin-grid ObjectGrid.tsx fieldsToShow (uncapped). Fixed at the producer: @object-ui/core gains resolveNameField (name-space twin of getRecordDisplayName; the declared nameField/displayNameField/NAME_FIELD_KEY chain now has ONE spelling read by both, so they cannot drift) and leadWithNameField, called by all three faces. Chose core over plugin-detail's resolveTitleField because that one runs a different ladder and is not a plugin-grid dependency. The name field is MOVED to the front, not merely appended, and on the two capped faces the lead is applied BEFORE the slice. Deliberately untouched: author-declared columns/fields lists (both pinned as control tests), ObjectGrid's uncapped no-highlightFields walk (the defect is unreachable there — nothing can fall off the end), and the showcase app in objectstack. Three cases decline to lead: no field def, hidden:true, and a DERIVED pick landing on a system-managed column (objectui#2702/#2777); a DECLARED nameField on a system field still leads (sys_migration points at id).",
 "tests": "Union re-run on final merged head 34b307b38. LIVE END-TO-END (the reported surface): this branch's source on a Vite dev server :5181 proxying to the showcase backend :3911, /apps/showcase_app/showcase_account view 所有记录 — thead th = ['', '#', '客户名称', '生命周期', '行业', '年收入', '操作'], tbody tr = 14, row 1 = ['', '1', 'Vandelay Industries', 'Active', 'Finance', '6,700,000']. Before (per the card): #, 生命周期, 行业, 年收入, 操作 — no name column. My dev server stopped by recorded PID (40662 + child 40666); :5181 confirmed free; PM's :3911 untouched. REVERSE VERIFICATION: ablated the three call sites to origin/main keeping the core helper and every test, so faces fail on assertions not a missing import. Predicted red; measured 10 failed / 46 passed, every failure inside a #7245 block and every pre-existing assertion in those files still green. Mutation proven on disk BEFORE measuring (leadWithNameField occurrences 4/4/3 -> 0/0/0 and each file's blob hash equal to its origin/main blob); restore proven by OBSERVATION after (git diff HEAD empty, git status clean, all three blob hashes byte-identical to the pre-ablation reading), not by an exit code; script carried a trap with absolute paths and treated an empty hash as failure. NO rebuild leg needed and none claimed: vitest.config.mts:277 aliases @object-ui/core to packages/core/src, so nothing resolves through dist. GATES: targeted vitest 14 files (both new suites, both edited suites, blast radius, hostFetchedDefaultColumns-6677, plus the just-landed ListView.speculativeFls-7216) = 'Test Files 14 passed (14)' / 'Tests 207 passed (207)'. Full packages/core/ + packages/plugin-grid/ = 'Test Files 1 failed | 217 passed (218)' / 'Tests 1 failed | 3139 passed (3140)' — the one red is rowRecordCrudVerdict.test.tsx 'paginates under the server's cap' with 'Error: Test timed out in 15000ms' at 16471ms, i.e. a TIMEOUT not an assertion, the flake signature AGENTS.md documents; re-run alone 'Test Files 1 passed (1)' / 'Tests 15 passed (15)', and the file covers batched CRUD verdicts, not column synthesis. type-check on core/app-shell/plugin-grid all 'Done', each echoing 'tsc --noEmit && tsc -p tsconfig.test.json' (checked the script name really ran — no zero-match silent pass — and the second invocation means the new test files ARE type-checked). eslint over the three packages: 0 errors, 1397 files processed (count from --format json). check-changeset-presence.mjs: '8 source file(s) of 3 released package(s) changed, and this change declares 1 changeset(s)'. check-changeset-no-major.mjs: 'No changeset declares a major bump.' TWO DECLARED NARROWINGS, both also in the PR body: (1) os-verify-lock.sh found no usable flock on this macOS host and ran every command in DECLARED UNLOCKED mode — its own official wording is pasted in the PR body; nothing was serialized against sibling agents. (2) 'pnpm exec vitest run packages/app-shell/' was SIGTERM-killed by the ~10-minute foreground cap (exit 143) before any verdict, and was replaced by a 10-file blast radius (every app-shell test referencing defaultListColumnsFromObject / defaultColumnsFromObject / getRecordDisplayName / deriveTitleField, plus all ObjectView.* and InterfaceListPage.* suites) = 'Test Files 10 passed (10)' / 'Tests 110 passed (110)'. The rest of app-shell is left to CI. LINT SCOPE EVIDENCE, all three: population read from eslint's own resolution of the three package dirs under the root flat config; count 1397 from --format json; invariance — eslint.config.js declares no parserOptions.project and no projectService, so type-aware linting is OFF and this diff, which changes no config file, cannot move any untouched file's verdict.",
 "mcp_calls": "0 — no MCP GitHub calls; the repo-scoped REST channel (gh) was probed first and worked for every read and write, including the dedupe search",
 "open_questions": [],
 "out_of_scope_findings": [
 "filed as #7287 (unassigned, labels finding + domain:ui): plugin-detail's resolveTitleField reads `primaryField` — a DetailViewSchema key (packages/types/src/views.ts:537, routed as a LOCAL unmirrored DetailView key by zod-mirror-parity.test.ts:894, absent from @objectstack/spec's shipped dist entirely) — off an OBJECT definition, ranked ABOVE the canonical nameField. Same shape objectui#6531/#6557 removed for objectDef.titleField. It also runs a second, divergent fallback (a five-name literal list vs core's type-aware deriveTitleField), so the detail page's H1 and every other surface can title the same record differently. Deduped first: REST search for resolveTitleField (0), primaryField (2, both about the mirror-parity ledger), 'name field ladder' (3, unrelated); control word nameField hit #7245 itself, so the search was live."
 ]
}

[hotlong]

ACCEPT — reviewer of record: PM session session_62849a16-0144-4728-8942-9f60bb3a73f1.

Checked against GitHub, not the report: draft PR #7284 targets main, first line Fixes #7245, no other closing keyword, changed files stay inside the claimed surface (no governed path), changeset declared, every check green at the current head. Spot-read: core/app-shell/plugin-grid lead the synthesized default list with the object's name field — 3 synthesis faces, producer-side helper in @object-ui/core, ablation 10 red/46 green, live e2e on the showcase Account list (name column first), changeset present. Flipping to ready and arming squash auto-merge; the queue lands it.

PR: #7284

Generated by Claude Code