RED-first adversarial guards pinning machine-proven fail-first BEFORE any
producer change. They inject a new proveFailFirst option the current producer
does not read, so they FAIL against the attestation-greens code
(src/prohibition-enforcement.cts:411). This RED is INTENTIONAL — Plans 02-03
wire the prover and turn them green.
New tests:
- FF-01: attestation + clean pass but no machine proof -> hard-gate (currently
RED: producer greens on attestation; actual 'green' vs notStrictEqual 'green')
- FF-01 positive: proven-fail-first + clean pass -> green (passes now, pins shape)
- FF-04: passes-on-violation (prover could not prove red) -> hard-gate (RED)
- FF-04: fails-on-clean even when proven -> hard-gate (passes via runCheck:false)
No source edited. 2 new guards RED, 2 positive controls green.
Maintainer CHANGES_REQUESTED (reviewed e08667e5, pre-portability-fix):
- B1: lint-rule no longer greens an unparseable target (eslintHasFatalError -> fail
closed on any fatal/parse error) or an inline-suppressed violation (eslintJsonHasRule
now scans suppressedMessages too). RED-first + real-runner repros.
- B2: both child spawns get a bounded timeout (30s node / 60s eslint) + 16MiB maxBuffer;
timeout fails closed. Injectable timeoutMs (positive-only — 0/negative can't disable
the bound) enables a fast 1.5s hang test.
- M1: scoped verify-phase.md — the check descriptor is author-supplied for now; filed
#1278 for deterministic auto-locate of the descriptor (the locate half).
- M2: added tests/prohibition-enforcement.property.test.cjs (fast-check fail-closed
invariants; within the <=2-file budget).
- m1: tapTestNames excludes # SKIP/# TODO; parseNodeTestSummary tracks # cancelled;
isNonVacuousNodeTestPass requires cancelled===0.
- m2: scoped the determinism claim to the decision/parse layer (real runner is env-dependent).
- m3: filed #1279 for machine-proven fail-first (violation-fixture probe).
- n1: -- before target in both arg builders (option-injection). n2: dropped dead token.
- B3 (Windows npx) was already fixed in 2af76306 (pushed ~65s after the review).
Verified node 22 + 24; size baseline regenerated for the verify-phase note.
A mutant dropping baseOf() from the TARGET side of the node-test vacuity compare
survived — add the mirror case (relative TAP name vs absolute descriptor target,
both basename-normalized -> vacuous).
CI failed on node 24 + Windows in the real-runner E2E tests — real portability
bugs in the shipping runner, not test flakiness:
- node 24 names a zero-test file's TAP entry by ABSOLUTE path (node 22 used the
basename), so the exact-string vacuity discriminator misfired -> empty file
falsely greened. Now compares BASENAMES (separator-agnostic) -> robust across
OS + node version. Verified: empty file is non-green under node 24 locally.
- the lint-rule runner spawned 'npx', which execFileSync can't launch on Windows
(and failed fast on ubuntu-24). Now resolves the project's eslint CLI via
eslint/package.json and runs 'node <cli>' through process.execPath (portable;
no shell -> no injection). eslint absent -> fail-closed.
- node-test also spawns via process.execPath, not bare 'node'.
Also folds in round-3 WR-01: the probe-core inline comments still said
enforcement was 'deferred to a follow-up PR' — corrected (landed in #1259).
Strengthened the isNonVacuousNodeTestPass unit test to pin the basename clause
(absolute-path target). Verified 71/71 green under BOTH node 22 and node 24.
Round-2 adversarial review found the lint-rule runner falsely greened an
eslint-IGNORED target: eslint returns a length-1 "File ignored" result (ruleId
null) that passed the >=1-file vacuity guard while nothing was linted — reopening
the vacuous-green class. Fix: buildLintArgs now passes --no-warn-ignored so an
ignored path returns [] -> fails closed (verified + E2E test on an ignored
bin/lib artifact).
Also: wrap runCheck() so even a (test-injected) throwing runner fails closed
(NEW-WR-01, full no-throw contract); document the benign basename-naming
constraint on wired node-test names (NEW-WR-02, fail-closed).
Adversarial pre-submission review found the injected-runCheck tests masked a
non-functional real runner. Fixes:
- BL-01 (false green on vacuous test): the node-test runner now parses the TAP
summary and requires a NON-VACUOUS pass (>=1 test, >=1 pass, 0 fail) AND a
reported test named distinctly from the file — node --test counts an empty file
as one passing test, so counts alone could not catch it.
- SF-01 (lint anchor never greened): the lint-rule runner now runs the project
eslint as --format json and filters by ruleId, so plugin rules (local/*) load
via the flat config — bare --rule cannot load a plugin. local/no-source-grep
now genuinely greens (covered by a real, non-injected test).
- BL-02 (tautological fail-first): the runner no longer echoes the caller's
failFirst as if confirmed. failFirst is documented as caller-ATTESTED; the
producer requires attestation + a genuine non-vacuous pass. Machine-proven
fail-first (needs a violation fixture) is flagged as a tracked follow-up in
ADR-550, the changeset, FEATURES, the reference doc, and verify-phase.
- SF-02: added real-runner end-to-end tests (no injected runCheck) + pure,
exported parse/filter helpers (parseNodeTestSummary, tapTestNames,
eslintJsonHasRule, eslintFileResultCount) so the shipping branches are
mutation-pinned.
- NIT-01/02: LOCATE guard rejects empty-string rule and unknown kinds.
- Hardening: spawn checks with NODE_TEST_CONTEXT/NODE_OPTIONS scrubbed so an
ambient test-runner context cannot corrupt a verify-time result.
- Docs reconciled to the shipped behavior (no 'confirms fail-first' overclaim).
