[finding] sync-template-versions.mjs's GAP 1 note says Test Core is skipped in full for a diff confined to it — no longer true once the scripts filter lands

[claude]

Blocked-by: #10014 (only becomes true when that PR lands)

Found while implementing #9829. Filed unassigned.

scripts/sync-template-versions.mjs's --self-test header argues the flag's existence from two
measured gaps. GAP 1 (lines ~376-401) reads:

GAP 1 — SCHEDULING. create-objectstack#test is reached only from ci.yml's test job, and that
job is if: ... needs.filter.outputs.core != 'false'. […] a diff confined to this file yields
core=false, so Test Core is skipped in full and the vitest never runs, on precisely the PR that
changes the rewriter.

PR #10014 changes the second half. core=false stays true — that filter is untouched — but the
test job now also gates on a new scripts filter output, so on a diff confined to this file the
job runs, --union-into runs inside it, and create-objectstack is pulled into the shard's package
set. Measured on that branch:

$ node scripts/check-cross-package-test-inputs.mjs --union-into ls.json --changed changed.txt
Cross-package scans pulled into this run because the diff touched their declared inputs:
  + @objectstack/spec  (declared glob matched scripts/sync-template-versions.mjs)
  + create-objectstack  (declared glob matched scripts/sync-template-versions.mjs)

So after that PR the paragraph states a scheduling fact that is no longer the case, in a header whose
whole job is to justify why the self-test lives in lint.yml. Same shape as #9651: a rationale
comment that stopped being true and would have the next reader re-derive the wrong conclusion.

What is NOT affected, and why the flag itself should stay: GAP 2 (the red paths — every failure
contract in the script's header, the clean-corpus control, the problems collection) is untouched,
and lint.yml carries no paths filter at all, so it still runs on every pull request, push and
merge-queue build. The fix is a comment edit — record that the scheduling half was closed and by
what — not a relocation of the self-test.

Deliberately not folded into #10014: that PR's diff is confined to .github/workflows/ci.yml under a
minimal-diff ruling, and editing this file would pull check:template-version-sync into its gate
set for a comment.

Refs: #9829, #10014, #9348 (the card that added the flag), #9651 (same class).

---

Generated by Claude Code

---

Generated by Claude Code

[claude]

Claim: PM loop round 22
Session: session_01XqDQYVU5smx29ts9pAErja
Branch: claude/issue-10016-gap1-header-scheduling
Worktree: objectstack-issue-10016
Domain: domain:devx
File surface: scripts/sync-template-versions.mjs (the GAP 1 header) (stop on breach; explain in the report)
Container & model: S, mode:subagent, model: opus (node scripts/pm/dispatch-gates.mjs --tier scripts/sync-template-versions.mjs — content limb no, comment prose only)
Clause-②: no
Serial constraints cleared: swept all open PRs for any touching sync-template-versions — none. UNBLOCKED: the Blocked-by PR #10014 (#9829) is now MERGED, which is precisely what makes this card's premise true — the header's scheduling claim stopped being accurate the moment that landed.

---

Generated by Claude Code

[claude]

{
  "issue": 10016,
  "status": "done",
  "branch": "claude/issue-10016-gap1-header-scheduling",
  "pr": "https://github.com/objectstack-ai/objectstack/pull/10047",
  "premise_still_valid": true,
  "summary": "The GAP 1 header in scripts/sync-template-versions.mjs is rewritten to record that the scheduling gap CLOSED, with the mechanism attached, instead of asserting a fact ci.yml no longer supports. Both halves of the gap were measured separately per ruling 1, and BOTH turn out closed: (a) scheduling -- core=false still, but the test job now ORs in a second filter output (scripts: 'scripts/**'), so its if: resolves to RUN; (b) coverage -- the vitest that covers this script really does execute on such a diff, traced link by link and then confirmed end-to-end with turbo reporting 'Cached: 0 cached, 1 total'. The header therefore says GAP 1 no longer justifies the flag and GAP 2 (the red paths) carries it alone; it records GAP 1 rather than deleting it so the next reader does not re-derive the retired conclusion. Also corrected in the same paragraph: the header cited picomatch 4.0.5 as 'the matcher dorny/paths-filter uses' -- the action's own lockfile resolves 2.3.1. Diff is comment-only (48 insertions / 16 deletions, every changed line a // comment, verified mechanically); Clause-2 respected, no behaviour change, no changeset (skip-changeset label applied and read back).",
  "tests": "All at HEAD bf46ce42 (final commit). GATE UNION re-derived from the real changeset with `node scripts/pm/dispatch-gates.mjs` (no path args; it took the change set from merge base c1731d023 and named 1 path, 3 gate families). Gates, each quoted from its own verdict line, exit codes captured before any pipe: `pnpm check:template-version-sync` -> EXIT=0, '✓ sync-template-versions --self-test: 40 assertions over temp fixtures, running the real CLI.'; `pnpm check:cross-package-test-inputs` -> EXIT=0, 'All 52 self-test cases passed.' + 'OK: 12 package(s) read outside themselves, all declared, and turbo.json hashes every declared glob.'; `node scripts/check-cross-package-test-inputs.mjs` (the bare form ci.yml spells) -> EXIT=0, same OK line; `pnpm check:nul-bytes` -> EXIT=0, 'check-nul-bytes: OK (scanned 6358 text file(s) -- 6358 tracked, 0 untracked-not-ignored; skipped 5 binary; no raw ASCII control bytes).' H1 MEASUREMENT: parsed the filter YAML out of the merged .github/workflows/ci.yml (not transcribed) and evaluated it against changed=['scripts/sync-template-versions.mjs'] with picomatch 2.3.1 and {dot:true}, SOME quantifier, matching dorny/paths-filter@v4's createRuleItem -- 'picomatch version used: 2.3.1 / action: dorny/paths-filter@v4 / filter core => false (no packages/**, no examples/**, no apps/!(docs)/**, no package.json, no pnpm-lock.yaml, no tsconfig.json, no .github/workflows/ci.yml) / filter scripts => true (MATCH scripts/**)'. 2.3.1 is what dorny/paths-filter@v4's own package-lock.json resolves for picomatch (runtime, non-dev, ncc-bundled), fetched and read rather than recalled; the 4.0.5 in this tree agrees on these globs, cross-checked separately. So `!cancelled() && (core != 'false' || scripts != 'false')` = false || true = RUN. H2 TRACE (each link run, not inferred): `turbo ls --affected` with TURBO_SCM_BASE=origin/main -> 'packageCount: 0, items: []' on turbo 2.10.10, so the affected set alone tests nothing; `node scripts/check-cross-package-test-inputs.mjs --union-into ls.json --changed changed.txt` -> '+ @objectstack/spec  (declared glob matched scripts/sync-template-versions.mjs)' and '+ create-objectstack  (declared glob matched scripts/sync-template-versions.mjs)'; `node scripts/partition-test-shards.mjs ls.json --shard 2/3 --exclude @objectstack/dogfood` -> 'shard 2/3: 1/2 packages, weight 7' naming create-objectstack (spec lands on 1/3, 3/3 empty); turbo cache -- create-objectstack#test declares $TURBO_ROOT$/scripts/sync-template-versions.mjs in turbo.json, and `turbo run test --filter=create-objectstack --dry=json` gave hash fdb9dd70563b8fb5 at HEAD vs 9fa6c426413e2f96 with a one-line edit to that script, so no cached green can be replayed; create-objectstack's test script is a bare `vitest run` whose config includes src/**/*.test.ts. END-TO-END, the shard's own command under the shared verify lock (flock -E 99 -w 540 /tmp/os-heavy-verify.lock, acquired first try): `pnpm turbo run test --filter=create-objectstack --concurrency=2` -> '✓ src/template-version-stamps.test.ts (8 tests) 338ms', 'Test Files 7 passed (7)', 'Tests 81 passed (81)', 'Tasks: 1 successful, 1 total', 'Cached: 0 cached, 1 total'. That last line is the load-bearing one: the task executed rather than replaying cache. NO ABLATION was run -- this PR changes no executable line, so there is nothing whose removal could move a diagnostic; the reverse-verification here is the picomatch/filter evaluation itself, which was run against the merged workflow rather than against a claim. COMMENT-ONLY PROOF: `git diff -U0 | grep -E '^[+-]' | grep -v '^[+-][+-]' | grep -vE '^[+-]// ' | grep -vE '^[+-]//$'` matched nothing (exit 1); `node --check` passes; control-byte scan `grep -naP '[\\x00-\\x08\\x0b\\x0c\\x0e-\\x1f\\x7f]'` clean on the edited file and on the PR body.",
  "open_questions": [],
  "out_of_scope_findings": [
    "filed as #10045: .github/workflows/lint.yml (~536-556), the comment above the 'Template version-time rewriter self-test' step, is a second near-verbatim carrier of the SAME retired claim ('Test Core is skipped ENTIRELY ... and with it the vitest', '--union-into ... is a step inside the skipped job', 'The queue build catches it, one merge late') plus the same wrong picomatch version. Reported, not edited: ruling 3 reserves ci.yml and the gate wiring for the change that just landed there, and the claim's declared file surface is scripts/sync-template-versions.mjs.",
    "filed as #10046 (finding): check-cross-package-test-inputs.mjs's --union-into appends to packages.items but never updates packages.count, so the turbo-ls document it hands partition-test-shards.mjs contradicts itself (measured: count 0 alongside two items). Inert today -- partition-test-shards.mjs reads items and nothing in either script reads count -- but that document is the input to a shape assertion whose stated purpose is to stop a silently empty shard."
  ]
}

---

Generated by Claude Code