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
LAB: CRUX harness runs llama-cli under a controlling tty — 0-byte output with rc 0 #5197
Defect.scripts/crux_inference_dogfood.sh:704-705 launches the llama-cli cell with no setsid. When the run inherits a controlling tty, llama-cli writes to /dev/tty: the captured stdout and stderr are 0 B with rc 0, and the cell comes out blank. Probes: </dev/null was OK, a pty reproduced the 0 B, and setsid -w was OK. A run under setsid -w produced 1062 B with the expected answer.
Fix shape. Run the cell under setsid -w. Treat 0 B of stdout with rc 0 as a harness error, never a blank cell. Add a test that runs under a pty.
Owner: aprender-cd
LAB (C324): runs off the PR, merge-queue and release paths. A red updates this one ticket under its one owner and stays silent otherwise; red or unmeasured for 7 days, it stops until the owner brings it back.
Agent: aprender-a3 (drafted by aprender-w71-e3a; filed after v0.71.1 shipped)
Folded into #4715 as a checklist row (TRIAGE-100, 2026-10-14): the open-ticket cap is 100, so this ticket is now tracked on that master and closed as not planned. Nothing is lost: reopen it when its work resumes, and an open PR for it reopens it.
Activity
noahgift commentedon Oct 14, 2026
Folded into #4715 as a checklist row (TRIAGE-100, 2026-10-14): the open-ticket cap is 100, so this ticket is now tracked on that master and closed as not planned. Nothing is lost: reopen it when its work resumes, and an open PR for it reopens it.
Agent: aprender-clerk