B-JAP6-root-test-hop-pinning-self-location-times-out-under-load
State: promoted/task
The failing test is
T-XTGT self-location hop pinning > info-report's RUNTIME fallback self-locates via its real ×4 hop,
in solutions/ontological/lib/tests/hop_pinning.test.ts. It failed by TIMEOUT at 5000.13 ms — bun’s
5 s default, to the hundredth — not by assertion.
Load-sensitive by construction: the test deletes CLAUDE_PLUGIN_ROOT, imports
../services/plugin/info-report.ts and calls buildReport(), so it does real filesystem and
subprocess work inside a fixed default timeout.
Observations: PR #1722’s pre-merge affected sweep saw
root:test 2815 pass / 1 fail, then two reruns of the same task pass 2816 / 0; the post-merge sweep
saw 2814 pass / 1 skip / 2 fail (moon suppressed the per-test detail); a direct re-run of the suite
command
(bun test solutions/ontological scripts --path-ignore-patterns='**/target/**' --path-ignore-patterns='**/worktrees/**')
at load 11 reproduced it as 2815 pass / 1 skip / 1 fail with the name above.
Unrelated to that PR: on the branch, a name-only diff of origin/main...HEAD limited to
solutions/ontological is empty.
Suggested direction for whoever picks it up, stated as a suggestion not a conclusion: give the test
an explicit timeout sized for a cold subprocess, or hoist the buildReport() work out of the timed
body — the same shape of fix the Shiki flake needed, except here the timeout signature is
unambiguous.