Files

1.0 KiB

Row 197 proof observations (Sage)

Kept here, not in Herb's BUILD.md, because that file is one of row 197's 37 paths and the landing check (mos-1174, mos-1185 item 5) requires them equal to c45ddeb6b.

Open observation (mos-1185 item 3), 2026-10-21

Proofs 4 and 5 each stopped on a test outside 197's paths, in a pair wave, and both times in lane b:

  • Proof 4, pair1-b, pkg-conversation, K12 (cohort.test.mjs:292): populated undefined. Row 250, Vikunja #450.
  • Proof 5, pair2-b, pkg-bus, T27 (host-run.test.mjs:469): no identities right after ssh-add -t 1. Vikunja #453.

Mos checked the step log birth times: the twin step started in both lanes within half a second each time (proof 4 pkg-conversation a 11:03:59.343, b .855 CDT; proof 5 pkg-bus a 12:07:25.354, b .573). Waves run in lockstep, so that holds for every pair wave, the passing ones too. Lane b is the twin that starts later. Mos's reading: contention exposing fragile tests, not a 197 defect. Paired lanes add contention, and that is 197's cost, covered by #451.