fix(tooling): re-tally @objectstack/rest's TEST_DEBT composition from tsc

[claude]

Fixes #10821

TEST_DEBT['@objectstack/rest'] carried a per-code tally that summed to 147 while
declaring it was counted at 153, with no remainder clause — and the entry's own
errors: field records 155. The tally is now taken from one tsc run against a built
workspace closure and sums to the recorded number exactly.

⛔ No digit was adjusted to make the old numbers agree. Every number below comes from a
single measurement; the old ones are quoted only as history.

The measurement

Closure built with the same command lint.yml runs before this gate, then the gate's own
re-measure, both under this container's shared heavy-verify lock:

turbo run build --filter=./packages/* --filter=./packages/*/*   → Tasks: 70 successful, 70 total (6m58s)
pnpm check:type-check-debt                                      → 32 ledger entries re-measured in 282.1s,
                                                                  1897 raw tsc errors, none above its recorded number

Per-code histogram for this entry, counted by the gate's own rule
(TSC_ERROR_LINE, TS6059 dropped — 0 dropped here):

code	count	what it is
TS2835	72	NodeNext extensions
TS7006	49	implicit any
TS2550	16	Array.prototype.at against a lib older than es2022 (all 16 are that one message)
TS2554	14	wrong arity (all 14 are Expected 2-5 arguments, but got 1)
TS2769	2	singleton tail
TS2345	1	singleton tail
TS6133	1	singleton tail

72 + 49 + 16 + 14 + 4 = 155, which is errors: 155 exactly. The remainder is spelled
as the house form the siblings use (plugin-approvals, objectql): "plus 4 singletons
(TS2769 x2, TS2345, TS6133)".

Both questions the note raised, answered

(a) the 147-vs-153 gap is retired unattributed, not invented. The per-class breakdown at
e8db1a230 was never retained, so the 6 cannot be recovered — and this ledger's own rule is
that a made-up attribution is worse than an admitted gap. They are retired with the tally
that carried them.

(b) 153-vs-155 was never one measurement disagreeing with itself. 153 was measured at
e8db1a230; RECORDED was later lowered onto a different measurement, 155 at 55da611
(#6939); the composition was never re-taken across that move. That is precisely what the
"not re-tallied by class at the 155 below" clause was saying out loud — the note was
honest, not sloppy.

Rescaling the old tally onto 155 would have been wrong in any case, which is why this is
a re-tally and not arithmetic: the shape moved in both directions — TS2835 67 → 72,
TS2550 10 → 16 and TS2554 13 → 14 rose while TS7006 57 → 49 fell — and two classes the
old tally never named (TS2769, TS2345) are in the pile now.

errors: 155 is unchanged

tsc reports exactly 155 for this entry today, so the shrink-only ceiling neither moves nor
needs to. @objectstack/rest does not appear in the run's surplus list; the three entries
that do are named under out-of-scope below. This is a note-only change — no ratchet
number moves in either direction.

Structural facts now in the note, all from the same run: the 72 TS2835 sit on 72 distinct
lines across 64 files but name only 11 distinct import targets, and 49 of the 72 are the
single ./rest-server import; every one of the 22 files carrying a TS7006 also carries a
TS2835, with no exceptions; src/rest.test.ts alone holds 38 of the 155 and no other file
is above x5. That replaces the old "they are one repair, not 124" with a measured
"they are 72 import-line repairs, not 121".

One declared change beyond the entry

The COMPOSITION docblock (#10722) used this entry as its worked example for why per-code
tallies are deliberately not summed:

//     partial by construction: plugin-rest's tally sums to 147 and says so in
//     the same breath ("composition as counted at 153"), so summing them would

Re-tallying the entry makes that sentence false about the tree, in a comment whose job is to
stop someone "fixing" the rule into summing tallies — the docblock's own stated
false-red hazard. The rule is untouched; only its worked example is marked retired, and
the entry is named correctly (@objectstack/rest, not plugin-rest — that mis-naming in
this comment is where the card's own title got the package name from). The gate's self-test
(47 semantic + 59 observation + 29 re-measure + 28 built-closure + 19 auto-lowering cases)
passes, and this entry's note opens with a per-code tally, so COMPOSITION still abstains on
it exactly as before.

Verification — all on 9fef66af

Gate floor derived with node scripts/pm/dispatch-gates.mjs --repo objectstack-ai/objectstack
(no hand-written path list; 1 path in the change set), all 8 families run plus check:nul-bytes:

gate	verdict line
check:type-check-debt	--re-measure: OK — 32 ledger entr(ies) re-measured in 282.1s, 1897 raw tsc error(s) total, none above its recorded number.
check:type-check-coverage	check-type-check-coverage: OK — 65/78 workspace packages type-checked …
check:cross-package-test-inputs	OK: 14 package(s) read outside themselves, all declared …
check:entry-guard	✓ check:entry-guard: 139 scripts/ file(s) — every entry guard goes through invoked-as.mjs …
check:parse-guard	✓ check:parse-guard: 138 scripts/ file(s) — every TypeScript parse goes through ts-parse.mjs.
check:pnpm-filter-targets	✓ check:pnpm-filter-targets: 120/148 --filter occurrence(s) across 25 file(s) resolve …
check-ci-filter-parity.mjs	OK: all 89 declared cross-package glob(s) (77 unique) are covered …
check-cross-package-test-inputs.mjs	OK: 14 package(s) read outside themselves, all declared …
check:nul-bytes	check-nul-bytes: OK (scanned 6448 text file(s) … no raw ASCII control bytes).

Declared narrowing — lint, measured rather than asserted. The repo-wide pnpm lint was
narrowed to the changed file, and the narrowing is a measurement:

1. Population read from eslint's own resolution, not a guess: the change set is 1 path
   (dispatch-gates, three-dot against the merge base), and eslint returned a result object
   for it rather than an ignore warning.
2. File count read from --format json: results.length === 1, errorCount 0,
   warningCount 0, suppressedMessages 0, exit 0.
3. Invariance for untouched files: this repo enables no type-aware linting for any file
   — grep -c projectService eslint.config.mjs → 0, no parserOptions carrying project,
   no project: key — so a one-file, comment-only diff cannot move the verdict on any file
   it does not contain. (eslint.config.mjs's own header states the same fact and cites a
   positive control for it.)

CI runs the full farm regardless; this is the cheap half, run early.

Out of scope — recorded only, left untouched

The same re-measure reports three entries sitting below their recorded ceiling. All are
left alone:

• @objectstack/plugin-auth — records 109, tsc reports 97 (−12). This is [finding] @objectstack/plugin-auth's TEST_DEBT records 109 where tsc now reports 98 — an 11-error surplus opened by PR #10614 #10615's entry
  and was explicitly fenced off for this card. Untouched.
• @objectstack/plugin-approvals — records 348, tsc reports 347 (−1).
• @objectstack/runtime — records 227, tsc reports 226 (−1).

The gate prints all three on every run and #6376 is their standing tracking card, so no new
issue was filed for them.

Notes for review

• The histogram was produced from the gate's own measurement world, not a re-derived one:
  the script was temporarily instrumented to dump raw tsc output, the dump was histogrammed,
  and the script was then restored byte-identically (sha256 5a795e74… before and after,
  git diff --quiet exit 0, zero instrumentation tokens left) before the real edit was
  written. No instrumentation appears in this diff.
• No changeset: this edits a CI-internal script and releases nothing, which lint.yml's own
  changeset-family block names as the textbook skip-changeset case. The label is applied.

---

Generated by Claude Code