You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Four tests fail on main on the workstation (WSL) regardless of change (host baseline): fm-remote-reply, fm-remote-secondmate-lifecycle-e2e, fm-remote-job, fm-public-followup #7
Fork-local, never upstream (it is our host environment, not upstream code).
Baseline run 2026-09-08 00:15-00:35 EDT on the workstation (WSL), main at f0e4f36, box idle (load ~1), each script run serially with nice -n 10 timeout 900 bash tests/<t>.test.sh:
test
main (idle)
note
fm-remote-reply
FAIL (47s)
fm-remote-secondmate-lifecycle-e2e
FAIL (676s)
not ok - fleet snapshot did not select the remote structured-home projection
fm-remote-job
FAIL (34s)
not ok - an orphaned command mutated after worker crash recovery
these seven failed only in the no-mistakes test round that ran under load ~13 with several crew pipelines active; load flakes, not defects
Context: surfaced by the lavish-axi removal pipeline (run 01M1YVA8R5WXGE2EW9347E2C6Q, branch fm/drop-lavish-bootstrap-check) whose full-suite round reported 11 failures. Its diff touches bin/fm-bootstrap.sh, tests/fm-bootstrap.test.sh, tests/fm-session-start.test.sh, tests/lib.sh (umask 022), docs and AGENTS.md only; none of the four failing scripts exercise that code.
Logs: /tmp scratch on the workstation (WSL) (session 596ac745), summary lavish-baseline/summary.txt.
To do: reproduce each of the four on the laptop or the server host to tell host quirk (WSL, umask 0002 daemon, no remote fixture) from a real regression; fix or mark expected-skip per host.
Activity