[objectstack-fleet]

Requested by Tesla · project thread

Part of #25712 (4b PR 2e: the liveness ledger, card item 1; Tier S).
Part of objectstack#25677.

Clause-②: no
Metadata-protocol: none

维护者速读(草稿)

• 改了什么:
  • liveness 账本里写成「objectui @某提交: 路径」的 374 个证据指针(164 行)改为仓内路径 ui/<路径>,去掉 objectui 标记和钉的提交。verifiedAt 一律不动。
  • 107 行的 evidenceScope 由 cross-repo 改为 in-repo。同时引用 cloud 的那 1 行保持 cross-repo,原本没声明 scope 的 56 行仍不声明。
  • 读取器把 ui 当作本仓根目录,ui/ 路径和其他路径一样必须存在、行号在文件内、锚点符号存在、文件里出现该键名。
  • objectui 改为「已退役的 realm」:仍标成 objectui 的路径单独列出并判红,同时打印对应的 ui/ 写法。
• 为什么改: objectui 已并入本仓(objectui 并入 objectstack 为 ui/ workspace：D1 裁定（A+）的执行父单 — 一个 PR 同时改 spec 与渲染器，删除两仓互钉，@object-ui 保留包名并与 @objectstack 同一版本组，objectui 仓归档为迁移说明 #25677),这些证据不再是跨仓的。
• 行号怎么处理(PM 已裁定): objectui 的 133 个 路径:行号 是在旧提交上读的。对照今天的 ui/:6 个行号未变,66 个代码挪了位置,58 个无法核对,3 个所在文件已删除。规则如下:
  • 只有当整段引文在旧提交和今天的文件里都恰好出现一次、逐字节相同时,才改到新行号。
  • 其余去掉指针上的行号,保留路径和锚点,旧读数留作文字 (:N at objectui @sha)。
  • 同一句里的行号要么一起改,要么一起保留。
• 读者已消失的 5 行: 删去失效的那一句,加一条带日期的 note,不猜新路径。其中 field.externalId 唯一的读者已被删除,记为后续 F23。
• 新增 13 条 key-mention 豁免: 这些 ui/ 文件读的是改名后的键(如 inlineEdit → editable)或委托给已锚定的兄弟文件,每条都写明文件实际用的名字。
• 风险与代价(含回滚): 只改账本、读取器、测试和 README,不影响运行时。回滚 = revert 本 PR。
• 你要做的: 无(Tier S)。

席位核对

(left for the dispatch seat)

Why

objectui merged into this repo as ui/ (#25677). The liveness ledger still cited the renderer as another repository: objectui @<sha>: packages/… pointers, counted foreign, never resolved. Card #25712 item 1 asks for:

• the pointers rewritten in-repo (ui/<path>#<symbol>);
• the scope set to in-repo;
• the reader to resolve them exist-or-fail.

PM ruling A (#25712 comment 6415397254) sets done-when to 0 objectui-attributed pointers, enforced by the reader. Note history stays. Unresolved rows go to the worklist, and no path is guessed.

What changes

Reader: scripts/liveness/evidence.mts, check-liveness.mts

• ui joins REPO_ROOTS. A ui/… pointer is local, so it gets the existence, line-bound, symbol-anchor and key-mention checks.
• objectui leaves FOREIGN_REALMS (now cloud, ee) and becomes the one entry of RETIRED_REALMS (objectui → ui). scanEvidence returns a third bucket, retired.
  • The gate fails on a path in it, at any status and in evidence or producer.
  • The message prints the ui/ spelling: query/limit → objectui: packages/core/src/index.ts (now ui/packages/core/src/index.ts).
  • Neither existing bucket would do. Counted foreign, the path never fails. Read as local, objectui: packages/core/… could pass against this repo's own packages/core/, a file that never was the reader.
• A path already spelled ui/… is local whatever prose precedes it, so a dated reading can still name the old repo.
• empty-state.mts and the key-mention.mts remedy text say cloud: where they said objectui:.

Ledger: 24 files

	Count
Rows with an objectui-attributed pointer in evidence or producer	164
Pointers re-rooted under ui/	374
objectui [@sha]: markers in front of a pointer, dropped	347
objectui words left as prose (no path directly after)	50
evidenceScope cross-repo → in-repo	107
Stays cross-repo (also cites cloud)	1
Scope undeclared, left undeclared	56
verifiedAt changed	0

Untouched:

• note history, including the "read at objectui @sha" readings and _note fields;
• 13 evidence strings that name objectui in prose with no repo-rooted path (objectui action renderers + runtime dispatcher);
• every cloud row.

Line citations: the 133-citation breakdown

Each objectui path:N was read at a pinned objectui commit. A clone of objectui (unshallowed) was compared with today's ui/ at this PR's base:

Measured per line	Count
Same lines today	6
Moved: the cited block is byte-identical at exactly one other offset	66
Unverifiable: the pin is not in objectui main's history (e.g. 940ba24, 33 lines), or the cited lines changed	58
In a deleted file	3

Rule applied (PM-accepted):

• Exact relocation only. A block must occur exactly once in the file at the cited commit and exactly once in today's file.
• One clause, one reading. A pointer moves together with its bare :N follow-ons (…ListView.tsx:763 seeds … and :1491-1494 filters …), or none of them moves.

Pointer lines, as applied	Count
Kept (same)	7
Relocated	58
Line dropped from the pointer; reading kept as prose ui/p (:N at objectui @sha)	65
In a deleted file: clause dropped or turned to prose, see below	3

• Follow-ons. There are 50 follow-on :N refs: 18 relocated with their clause, and 32 stay at the dated reading.
• Prose line numbers. Numbers written as prose inside a relocated clause were relocated by the same rule, in 8 rows:
  • sibling action-icon.tsx:N refs in action.bodyShape and bodyExtra;
  • :N/:N pairs in action.method and view.form.subforms;
  • "line N" in email_template.label/description and webhook.label/description.
• Rebases. Each time main moved, the relocation was regenerated from main's ledger by the same script, not hand-edited:
  • fix(plugin-list,plugin-grid): list empty state tells no-access and "mine" apart from no records #25987 moved ListView.tsx by 9 lines; view.list.hiddenFields now cites :2020, :3812-3815 and :3830;
  • feat(layout): a collapsed sidebar nav group shows its reachable child count #26103 moved NavigationRenderer.tsx by 28 lines; app.navigation.runAction's producer now cites :850, :870-875, :963 and :968.
• Exact matches only. Nothing was relocated by similarity.

Rows whose objectui reader is gone

Each row below gets no repoint and a dated note line saying what left and when:

Row	Dropped clause	Why
app.navigation.id	UnifiedSidebar.tsx#useNavOrder	objectui#12059 made the sidebar menu non-reorderable; the other three readers stay
view.list.hiddenFields	spec-bridge list-view.ts:123	objectui#6632 retired the spec-bridge; the ListView reader stays
view.form.groups	spec-bridge form-view.ts:169	objectui#6632; the producer fold in view.zod.ts and the ObjectForm reading stay
field.widget	plugin-detail/src/index.tsx:563 (permission-facet-link registration)	objectui#12998 removed the widget; found by the key-mention check once the path was in-repo
field.externalId	its only pointer, console metadataConverters.ts:125	deleted in objectui dc14bcb ("zero importers"); the pointer is now prose, F23

Key mention: 13 exemptions

Once in-repo, these citations met the key-mention check for the first time. Each cited file reads a renamed key, or names the parent key and delegates to the entry's other, anchored citation. Each row in key-mention.baseline.json names the spelling the file actually uses:

Entry	File	Name used
page/print.{paperSize, orientation, margins.right, margins.bottom, pageNumbers, repeatTableHeaders, avoidBreakInside} (7)	layout/page.tsx	print (isPagePrintDeclaration(schema.print))
dashboard/widgets.chartConfig.{showDataLabels, annotations, interaction} (3)	plugin-dashboard/src/DatasetWidget.tsx	chartConfigPresentation
view/list.inlineEdit	plugin-grid/src/ObjectGrid.tsx	editable
view/form.subforms	plugin-form/src/MasterDetailForm.tsx	details
translation/flows.screens	app-shell/src/views/ScreenView.tsx	label / placeholder / inlineHelpText

The baseline stays shrink-only. A row fails once its pair anchors.

Tests and README

• evidence.test.ts: the realm cases move from objectui to cloud/ee, and a new block pins the retired realm. Its ui/ paths are synthetic (ui/packages/layout/src/x.tsx): the scanner never opens them, and a real path there would be an undeclared cross-package input to check:cross-package-test-inputs.
• check-liveness.test.ts:
  • the realm cases move from objectui to cloud/ee;
  • two subprocess cases fail on a retired-realm pointer in evidence, in producer, and on a dead row;
  • the summary-line pin reads (cloud / ee — not resolvable here).
• Restored. The path:NNN contract test and the line-citation floor come back. [finding] field.conditionalRequired is ledgered live citing a .transform that protocol 17 removed — and the rule-validator fallback its note relies on is gone too #13043 retired both when that population reached zero, with the instruction to restore them when a citation was written again.
• README.md updates the realm-marker section, the commit-pin advice, evidenceScope, the preview lookup (git grep -- ui/ instead of git -C ../objectui) and "confirmed absence".

The first commit is the governed-queue-guard comment fix held from #26058, accepted by the PM to ride here.

Follow-ups (sub-issue 4)

• F23 (new). Find the current reader of field.externalId, or re-classify it. Its only cited reader was a dead module.

• F24 (new, optional). The 65 relocated ui/ line citations are now in-repo path:N pointers, and they rot in range the way [finding] A liveness citation can rot IN RANGE: action.target and action.requiredPermissions point at the wrong lines one day after being repointed, and every check stays green #12516 describes: the two base moves above shifted seven of them in two hours. Converting them to #symbol anchors is the durable form.

• F25 (new, noted). docs/qa/platform-checklist/areas/*.json still write objectui: packages/… refs. That surface is not the ledger and is outside this card item.

• F26 (new, reported to the dispatch seat). After this PR, the gate and evidence.test.ts read 113 ui/ files through the ledger. A ui/-only PR schedules neither:

  • Spec Liveness Check's PR paths: do not list ui/**, and it is not a required context;
  • Test Core does not start, and @objectstack/spec#test:repo does not hash ui/**, so in the merge queue it is a cache hit.

  So a ui/ PR that renames a cited file or symbol lands green, and the next spec PR goes red (main is red for every PR that touches packages/spec: #7769 gave sys_api_key update without bulk, and the conformance scan that catches it lives in a package #7769 never touched #7802's shape). packages/** citations do not have this gap, because spec declares those roots. Closing it means declaring the ui/ sources in scripts/cross-package-test-inputs.mjs and turbo.json, and covering them in root ci.yml crosspkg, which check:ci-filter-parity requires. That is Tier H, and every ui/ PR would then start Test Core for spec. Left to the dispatch seat; not widened here.

Verification

Head 23cb18d2b9, on main 1bbb7bf490. 33 files, +703 / −385. Tier S: no governed path, under 3000 lines.

• pnpm --filter @objectstack/spec check:liveness, exit 0:

  • evidence paths: 1036 repo-local paths declared, 1036 resolved; 41 attributed to another repo (cloud / ee).
  • line citations: 65 written path:NNN, 65 inside the cited file.
  • symbol anchors: 1220 written path#symbol, 1220 named.
  • key-mention anchoring: 1036 pairs, 1022 anchored, 14 exempt.
  • Retired-realm pointers: 0.

• vitest run scripts/liveness/: 12 files, 358 tests passed.

• Gate sweep: node scripts/pm/dispatch-gates.mjs --commands derives 68 gates for this diff. All 68 ran on 23cb18d2b9 and exited 0. --ran reconciles 68 derived, 68 run, 0 NOT-MEASURED, 0 UNRUN.

  • The first pass found check:cross-package-test-inputs red on four real ui/ paths in evidence.test.ts. Those unit cases now use synthetic paths (see Tests).

• Push battery (GITHUB_ACTIONS=true GITHUB_EVENT_NAME=push node scripts/pm/check-dispatch-gates.mjs) on 23cb18d2b9: 2182 cases pass, 0 fail (fast 1917, slow 265; 1800 s).

🤖 Generated with Claude Code

https://claude.ai/code/session_01A6SEitNngAMugGbuJHWsDS

[github-actions]

📓 Docs Drift Check

⚠️ 25 changed file(s) yielded no anchor (packages/spec/liveness/README.md, packages/spec/liveness/action.json, packages/spec/liveness/app.json, …), so the pages documenting them are NOT COVERED by this run — this is not a clean bill of health for those files. Nothing else in this diff resolved to a documentable surface (no symbol, route or SDK anchor derived from 1 changed package(s)).

What this run could not see

• 25 changed file(s) yielded no anchor (packages/spec/liveness/README.md, packages/spec/liveness/action.json, packages/spec/liveness/app.json, …) — pages documenting those are invisible to this run
• a page that states a rule by its inputs shares no identifier with the emitter that implements the rule, so an emitter-only diff cannot list it — not on this run and not on any run. Measured on fix(driver-sql): emit varchar(maxLength) for a text field a declared index keys on #11430: content/docs/protocol/objectql/types.mdx documents the text-family column mapping by the ObjectQL type names it maps FROM (text / textarea / html) while the diff changed createColumn; it went unlisted, and it was the page that diff falsified, in four places. No shared token exists to detect this on, so a rule your change carries has to be re-read by hand in the pages that restate it.
• a key NAME is not a key, so the hand re-read the line above prescribes can land on the wrong schema. The same spelling is authorable on one governed type and a [REMOVED] tombstone on another for each of active, aria, joins, objects, template, tools and version (censused on [finding] tools is a key on BOTH AgentSchema (tombstoned, dead) and SkillSchema (live, cloud-attested), so a name-based search attributes skill examples to the agent key — it produced a false stop-the-line alarm on PR #19059 #19093 over the liveness ledger's governed types, top-level keys); nothing in a search result distinguishes the two, so a grep hit on a LIVE example reads as evidence about the DEAD key. Measured on fix(spec): the agent.tools liveness row says dead — it claimed live on a key the schema tombstoned #19059: content/docs/ai/agents.mdx was reported as contradicting the agent.tools tombstone over its tools: example at :161, which is inside the defineSkill({ block opened at :155 — the page was already correct. Settle ownership by PARSING the value against both schemas, never by the name: that literal PASSES SkillSchema, and as an AgentSchema it FAILS at tools with the tombstone prescription. ⛔ These names are not the whole class — a key retired through a .strict() guidance map leaves no tombstone in the walked shape and none of them here (tool.category, live as AIToolDefinition.category).

Coarse fallback — 145 page(s) merely mention a changed package (the pre-#9192 predicate, kept for the deliberately-wide backstop): node scripts/docs-audit/affected-docs.mjs --json a335f1defbea57ffe552d2e9efc5ad8d56a2e76a → packageMentionDocs.