Repository navigation
Conversation
…s python3, the explicit-test container has none (Refs #5806) The guard exits 2 'ENV - python3 missing' in the sovereign-ci container (workspace-test-shard, intel-clean-room-13, #5844 shard 5). guard_tree.sh already runs every scripts/check_*.sh on the host, where python3 exists, so the guard stays wired; the in-container .cmd could only ever fail. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
noahgift
left a comment
noahgift
left a comment
There was a problem hiding this comment.
Seat (non-author) bound to head 2f0438d. Receipt: handoff/triage-quorum-5860-2f0438dace.md. Opus APPROVE (executed); agy REQUEST_CHANGES but every item is 'unverified from the diff alone' (no tools), no defect named. Checks run at the head: guard_tree universe includes check_harness_server_cleanup.sh and plans it as run: (not skipped, host section, no container, no continue-on-error; ci.yml gate requires X86:guard-tree); check_explicit_test_commands.sh PASS (117 fragments); shard union 'union of 4 shards == pre-split list (117 commands)' is derived, not pinned; no other reference to fragment 817 or to the guard. Known, not fixed by this PR: the host-side 'apr stub' race on the same guard (#5854). Not unanimous in letter, so arming is a3/cop's call. Not armed by me.
Root cause of the intel-clean-room-13 failure on #5844 shard 5 (
check_harness_server_cleanup: ENV - python3 missing): ci/explicit-test-commands.d/817 runs the guard inside the sovereign-ci container, which has no python3 (scripts/check_shard_count_single_source.sh:6 documents it). The image did not change; the guard was mis-wired. guard_tree.sh runs every scripts/check_*.sh on the host, so coverage is kept.Separate, not fixed here: the same guard also BROKE on #5854 guard-tree (framework16-3): R row 'apr stub' polls descendants every 0.1 s for a short-lived stub_server.py apr and the harness exits first (race).
Not armed. Gate-path change: needs quorum + 2 non-author seats.
🤖 Generated with Claude Code