Row217 round2 test receipts
Sage CLEAR sage-4042/4043, 2026-10-24. Source revision 2d274a5b6. All runs
serialized; source files stayed unchanged through the whole-suite and mutant pass.
Commands used TMPDIR=/mnt/storage/scratch/tmp/tNoueZ and timeout -k 10.
- npm ci --ignore-scripts --no-audit --no-fund at the worktree root (300s bound), then packages/webui (300s). Actual directories, no canonical links.
- node --test packages/bus/tests/citation-availability.test.mjs (180s).
- python3 agents/beakley/work/queue-217/mutants.py M5-hide-library-unavailable (420s). Green control, the authorized pre-fix RED stand-in, restore+green.
- node --test --test-concurrency=1 with the seven new test files (900s): library/guard, business/reply-limits, cli/reply-limits, bus/citation-availability, bus/citation-guard, bus/citation-runtime and harness/citation-runner. Whole files.
- First attempt 56/57: new runtime fixture invalid-authority. Only that fixture was corrected to the established broker projection. runtime-fixture-retry.log is its 1/1 rerun (180s); touched-rerun.log is the whole-file 57/57 rerun.
- npm test in packages/library, then business (900s each), bus and cli (1200s each), harness and seat (900s each). Never overlapped package runs.
- python3 agents/beakley/work/queue-217/mutants.py (900s). The driver wraps each Node run in timeout -k 10 120. Each negative exits1, is preceded by a green control, restores its product bytes/hashes and then reruns the restored control green. No checkout product file is mutated.
summary.json holds complete counts; summary.txt is the short version. M1 anchors are recorded in m1-anchors.txt. The five -restored.sha256 files match current product files. Full outputs are retained, including the first new-test failure. Negative logs intentionally keep Node's whitespace-only diagnostic padding; standard git diff --check flags that padding, not source code. No binary logs. No old test failed or was edited. Sage runs scripts/test-.sh at landing.