2.2 KiB
Row 97 round 1 static review, verification pending
Candidate bdeca38dc8897b711b68f055cf346fc0ba6c04721e22720906c6b8fddc97902a, three paths, detached base 3bd3f85c. Reconstructed under ~/filbert-scratch/r97-review/wt with Researcher's build.patch applied by git apply --index and node_modules symlinked. Manifest 3/3 and no drift (static-receipt.json). No product source change is in the candidate.
The edge test races the first board request against five seconds, clears the timer, and reaches existing finally cleanup on failure. The held-board test intercepts only 10000 ms poll timers and accelerates them to 150 ms in the redirect test; other timer delays retain their behavior. It records board versus bus timers, checks no board timer with the machine toggle off including Pause/Resume and reload, and checks no new timer after sign-out even through forced Pause/Resume handlers. Those timer observations expose schedule mutants whose downstream guards otherwise suppress the network read. README names all six redirect requests and explains the accelerated ticks.
Static inspection found no blocking issue. Runtime checks still need to establish failure within 30 seconds, normal browser/server cleanup, timer mutant kills 13 and 27, and passing suites. Researcher's mutant runner kills the process group unconditionally even after normal exit, so an independent toggle-on run should check for leftovers before doing safety cleanup; that distinguishes test cleanup from runner cleanup. Mutant 14 is intentionally not claimed and appears redundant behind api's sign-out guard, subject to runtime confirmation.
The packet reports test-discord failed because its scratch lacked node_modules. Reconstructing with the standard node_modules symlink permits an independent dependency-complete check; this report does not treat that baseline failure as a candidate defect or claim it passed.
No browser or webui tests have been run in this review. Sage's priority is row 69, row 90 round 2 once requested, then row 97 and row 98. Other seats have reserved the test machine. Do a /proc cwd busy check across both home scratch and /mnt/storage/scratch before every test, and use process groups with timeout cleanup and source restoration for mutants. No verdict posted.