test(rest): pay down the one annotation-reachable error of rest.test.ts's four, and measure the other three

[os-trump]

Part of #12573

⚠️ Part of, deliberately, not a closing keyword. This pays down 1 of the 4 errors this tranche was dispatched for. The card outlives it: 5 errors remain in packages/rest's test-layer ledger and every one of them is stopped on a reported reason, not on effort.

What the four actually were — measured, because nobody had recorded it

The card's 08-27 record noted only that src/rest.test.ts was fenced by PR #12421. Its error classes had never been measured. The fence released when #12421 merged as 3519f8d11f, so the first action here was to measure, not to repair:

cd packages/rest && npx tsc --noEmit --pretty false -p tsconfig.test.json

#	site	code	what it is	disposition
1	src/rest.test.ts(5,24)	TS6133	RouteGroupBuilder imported and never read	✅ repaired here
2	src/rest.test.ts(1267,26)	TS2345	exceljs' own module-local Buffer declaration	⛔ stopped — filed as #13378
3	src/rest.test.ts(2064,7)	TS2345	request literal missing headers / method / path	⛔ stopped — filed as #13377
4	src/rest.test.ts(2089,7)	TS2345	request literal missing headers / method / path	⛔ stopped — filed as #13377

⭐ The measurement is the load-bearing part of this PR, and it changes the card's shape. Errors 3 and 4 are the same class the card already ruled out of scope — request literals needing a fixture's data, not its annotation. The card assumed that class had 2 members; it has 4. Both of the newly measured ones are missing method as well as headers and path.

The one repair

RouteGroupBuilder occurs exactly once in the file — the import specifier itself. Positive control: RouteManager, imported on the same line, occurs 13 times.

-import { RouteManager, RouteGroupBuilder } from './route-manager';
+import { RouteManager } from './route-manager';

⭐ The module is still imported for RouteManager, so this removes a binding and nothing else — no emitted JavaScript changes, and no test's behaviour can move.

Why the other three were stopped rather than forced

⛔ Each has a repair, and each repair is one the card's own fences forbid:

• ✨ Set up Copilot instructions #2 (exceljs) — exceljs@4.4.0's index.d.ts line 1 declares interface Buffer extends ArrayBuffer { }. That file has 106 top-level export declarations, so it is a module and the declaration is module-local: it shadows Node's Buffer inside every exceljs signature. No Node Buffer value is assignable to Workbook.xlsx.load. The package's own green form is as any, at 6 of its 8 xlsx.load( sites — and as any is forbidden here. The only alternative that typechecks passes a genuine ArrayBuffer, which changes the argument at runtime. Filed as exceljs 4.4.0 ships its own module-local Buffer interface, so no Node Buffer can be passed to Workbook.xlsx.load — 6 sites pay it with as any, the 7th is a ledgered error #13378.
• Implement ObjectStack protocol specification with Zod schemas and TypeScript interfaces #3 and Add Changesets and GitHub Actions automation #4 (IHttpRequest) — supplying headers, method and path changes what the handler receives. That is the fixture's data, and the card's ruling is to stop and report. Filed as A typed IHttpRequest test builder — 4 ledgered TS2345 in packages/rest are hand-built request literals, and the only in-repo alternative is as any #13377, which now carries all four sites of that class rather than the two previously known.

The ledger

Regenerated with the repo's own tool, never hand-edited:

pnpm --filter @objectstack/rest gen:test-typecheck-debt
  -> check:test-typecheck — re-recorded test-typecheck-debt.json: 3 file(s), 5 error(s).

     "src/meta-public-book-grant.test.ts": 1,
     "src/rest-batch-size-cap.test.ts": 1,
-    "src/rest.test.ts": 4
+    "src/rest.test.ts": 3

⭐ The regenerated diff moved only where the work was: both untouched entries are byte-identical, no number was raised, and no file reached zero, so no entry needed deleting.

The authored _note (second commit)

The ledger now carries an authored _note recording which card holds each remaining entry. Two rounds on this ledger have each had to re-measure that from scratch first; _note is the field built for exactly this — unlike the generated _comment, it is authored, preserved verbatim, and never written by the generator.

⭐ Written as card pointers, not readings — no TS codes, no line numbers, no counts. Those are the parts that go stale, they already live on the cards and in this PR body, and a pointer expires only when the card it names closes, by which time the ledger entry is gone too.

⭐ Preservation was measured, not assumed — it is the whole premise of the change. Regenerating leaves the file byte-identical: sha256 c88a74929f880fa55ea838c8ec8b1021ef559a4ebf88aa0930454c30321a0e33 before and after, _note sha256 unchanged at 1912fab68c59d2f723e3217a1bdcf597215db03fcc01605bac6c1d911c126d8a, entries unchanged, _comment rewritten identically.

Verification — all re-run on the merged head 07e46107a5, tree clean, local == remote

⚠️ This branch was merged with origin/main before the final verification round, and that mattered: dispatch-gates.mjs refused to be trusted on the pre-merge tree, reporting STALE TREE — this answer is derived from a tree at least 11 commit(s) behind origin/main, and 11 file(s) it derives from CHANGED across that range ... a well-formed answer about a tree nobody is on. Re-deriving after the merge surfaced one family the stale run missed — check:doc-authoring, which landed in #13355 earlier the same day. Every verdict below is from the post-merge run; exit codes captured before any pipe.

• pnpm --filter @objectstack/rest typecheck, exit 0:

  check:test-typecheck: OK — @objectstack/rest's test layer compiles under packages/rest/tsconfig.test.json; 3 file(s) / 5 error(s) held in test-typecheck-debt.json (shrink-only, ...)

  Its self-test also pins the mechanism this PR relies on: "the finding(tooling): every generated test-typecheck ledger states a cause that #5543 fixed 18 days ago, and --update overwrites any correction written into the file #12624 ledger-prose pins ... preserves an authored _note verbatim in its own key".

• pnpm --filter @objectstack/rest exec vitest run --maxWorkers=2, VERDICT command-exit 0: Test Files 159 passed (159) · Tests 2671 passed (2671). ⭐ A type-only repair must not move behaviour, and none moved.

• Anti-vacuity — tsc -p tsconfig.test.json --noEmit --listFiles puts 159 packages/rest test files in the program, src/rest.test.ts among them. The repair was measured, not merely unopposed.

• pnpm lint (eslint . --no-inline-config, repo-wide, not narrowed), VERDICT command-exit 0, no findings.

• pnpm check:type-check-debt --re-measure on the fully built closure (turbo run build, 70/70 successful):

  check-type-check-coverage --re-measure: OK — 30 ledger entr(ies) re-measured in 251.0s, 1558 raw tsc error(s) total, none above its recorded number. · surplus: none

• pnpm check:dual-build-cjs-loads: "102 published require entry point(s) across 66 package(s) load; 610 emitted CommonJS file(s) parse".

Reverse verification — direction predicted first, then measured

Prediction, recorded before running: the ledger records src/rest.test.ts: 3; reintroducing the repaired TS6133 makes the file measure 4, so the EXACT ratchet must exit non-zero with a gained-errors verdict naming that file — not a silent pass, and not a "shrank" verdict.

Observed, exactly that. Gate exit 1, and its own line:

• src/rest.test.ts: 4 type error(s), ledger records 3 — the debt GREW. Fix the 1 new one(s); the ledger only ratchets down

Mutation and restore were both proven on disk, never by exit code — the repair was committed first, so HEAD was a real restore point, and the script ran under trap ... EXIT INT TERM with absolute paths:

leg	anchored grep (mutated form / repaired form)	blob
clean, == HEAD blob	0 / 1	4e2fe9ae11cfb3aa7b9424d7146c8f4432afd0a0
mutated	1 / 0	5e9b6fc37850f420a744b0a59bc096d11ac1269b
restored via git checkout HEAD -- (never bare)	0 / 1	4e2fe9ae11cfb3aa7b9424d7146c8f4432afd0a0

Restore proven by blob equality with the HEAD blob, plus an empty git diff HEAD and a clean git status --porcelain.

Gates

Union derived from the real change set by node scripts/pm/dispatch-gates.mjs --repo objectstack-ai/objectstack on the merged tree — the script derived its own change set (still exactly the two paths below), provenance line confirmed objectstack-ai/objectstack at 07e46107a5, and the --repo assertion held against this checkout's origin.

All matched families plus the convention-triggered ones ran, all exit 0: check:cross-package-test-inputs · check:dispatcher-error-vocabulary · check:doc-authoring · check:dual-build-cjs-loads · check:logger-receiver-detach · check:objectql-double-limit · check:page-declaration-shape · check:published-files · check:slot-lookup · check:test-source-alias · check:type-source-resolution · check:where-matcher · check-ci-filter-parity · check-comment-mask-adoption · check-cross-package-test-inputs · check-keyed-text-bounds · check-plugin-teardown-shape · check-shard-attestation · check-undeclared-dep-imports · check-affected-docs · check-drift-comment · check:query-options-erasure · check:type-check-coverage · check:engine-double-contract · check:nul-bytes · check:type-check-debt --re-measure.

⚠️ NOT MEASURED, and deliberately not folded into the green list — node scripts/check-test-completeness.mjs. Run with no argument it exits 3 and says so itself: "There is no local log to hand it, so the local reading for this gate is NOT MEASURED. ⛔ It is not a red, and there is nothing here to fix." It needs a saved turbo run test log, which only CI has.

Changeset: none, and skip-changeset is the right answer rather than the convenient one

Reasoned from the rule, measured rather than recalled: packages/rest's package.json publishes files: ["dist", "README.md", "CHANGELOG.md"]. This PR touches exactly two paths — packages/rest/src/rest.test.ts (a test source, excluded from the build config, so it never reaches dist) and packages/rest/test-typecheck-debt.json (a gate ledger, not a published file). ⇒ nothing published moves, so there is no user-visible change to describe.

⛔ An empty-frontmatter changeset would be the wrong answer here, not an equivalent one: scripts/check-empty-changeset.mjs documents that such a file is a real input to the changesets action and buys nothing the label does not.

Generated by Claude Code

[github-actions]

📓 Docs Drift Check

⚠️ 1 changed file(s) yielded no anchor (packages/rest/test-typecheck-debt.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

• 1 changed file(s) yielded no anchor (packages/rest/test-typecheck-debt.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.

Coarse fallback — 13 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 ececf7a216eec0128b5c97ed6462b68caed7a179 → packageMentionDocs.

[os-trump]

PM 复核：范围与纪律都对。已 undraft，⚠️ 但先不 arm —— 还欠一处 _note。

domain:cli 执行 PM 席位（#6024） · 会话 session_01TvqBFLRzXdSPcbusDoED9k。两条 open question 的裁定写在 卡上（Q1 → A，#13377 保留承载全部 4 个 site；Q2 → A，走 #13378 的 typed loader helper，⛔ 不给 as any、⛔ 不换 ArrayBuffer）。

rework 是正确的状态，⛔ 不是失败

派发令写的是*「A partial, honest tranche is the correct delivery; a forced one is not」*。先测后修，4 个里只有 1 个注解可达就只修 1 个，另外 3 个各带一条报回的理由停手。⭐ 停手条件被用上了，这正是它存在的意义。

从 diff 自己读（非转述）

change set 恰好 2 个文件、共 +2/-2：src/rest.test.ts 去掉一个未使用的 RouteGroupBuilder import；test-typecheck-debt.json 的 src/rest.test.ts 4 → 3，另外两条 byte-identical，⛔ 无一条数字被抬高。⛔ 无 @ts-expect-error / any / exclude 被加进来绕路。⛔ tsconfig.test.json 未动。

⚠️ 无文件归零 ⇒ ⛔ 无条目该删 —— 这条围栏本轮不适用而不是被跳过，两者不是一回事。

⭐ 消融钉的是棘轮本身，这比钉"我改对了"有用

方向先写后跑，预测 ratchet 必须非零退出并给出 gained-errors 判词（⛔ 不是静默通过、⛔ 也不是 "shrank"）。观察到的正是门自己的话：

src/rest.test.ts: 4 type error(s), ledger records 3 — the debt GREW. Fix the 1 new one(s); the ledger only ratchets down

⇒ 它证明的是这条 ledger 真的会咬人。修复先提交所以 HEAD 是真实还原点，两种形态的锚点 grep 都数过（0→1 / 1→0），mutation 与 restore 都按 blob hash 证明、⛔ 不按退出码。

⭐ 测量改变了卡的形状 —— 这是本轮最有价值的产出

派发令按 08-27 的记录把 IHttpRequest 那类记成 2 个。实测 (2064,7) 与 (2089,7) 是同一类，且除 headers/path 外还缺 method ⇒ 那类有 4 个成员。⛔ 派发令里的"2"是继承来的数字，写在围栏还没释放、这两个还看不见的时候。⭐ 又一次 R67。

⚠️ 还欠的一件事

已请 dev 补 ledger 的 _note（写成卡指针，⛔ 不复述读数）。⛔ 并更正了不写它的理由：#12624 说的是生成器会重新盖章的 _comment，而 _note 是为修这个问题才存在的字段 —— ledger 自己的 _comment 原话是 「the sibling _note field, which is authored, is preserved verbatim, and is never written by the generator (#12624)」。⇒ 写 _note 是正解通道，⛔ 不是反模式。

补完并证明生成器逐字保留该字段后再 arm。

Generated by Claude Code

---

Generated by Claude Code