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
Intermittent phantom single-failure in Node/Bun runtime suites #275
Twice now a runtime suite has reported one failure with no identifiable failing test: Node once during #264's battery (2117/2118, nothing greppable, two immediate re-runs clean), Bun once during #273's (1 failure with no (fail) line, three re-runs 2166/0). CI green both times; both diffs doc-or-unrelated. Pattern suggests something genuinely intermittent in the runtime suites, not the diffs.
Tracking issue so observations accumulate here instead of PR bodies. When it next appears: capture the full runner output to a file before re-running, note runtime + version + suite + diff context. No action until it has a name.
Another concurrent-only phantom Node failure occurred while verifying PR #415 at commit b82121c on macOS arm64 / Node v26.5.1.
The exact deno task verify battery reported test:node exit 1 after 271.8s, but the verifier correctly printed only the earlier Deno failure in full and supplied no identifiable Node failing test. Without a code change, pnpm test:node passed alone: 2,708 tests, 403 suites, 0 failures in 200.3s.
The same battery had a separately identified Deno browser-bundle cleanup failure, now tracked as #417. This observation retains the Node phantom evidence here rather than conflating it with the named Deno failure.
Activity
taras commentedon Aug 9, 2026
Another concurrent-only phantom Node failure occurred while verifying PR #415 at commit
b82121con macOS arm64 / Node v26.5.1.The exact
deno task verifybattery reportedtest:nodeexit 1 after 271.8s, but the verifier correctly printed only the earlier Deno failure in full and supplied no identifiable Node failing test. Without a code change,pnpm test:nodepassed alone: 2,708 tests, 403 suites, 0 failures in 200.3s.The same battery had a separately identified Deno browser-bundle cleanup failure, now tracked as #417. This observation retains the Node phantom evidence here rather than conflating it with the named Deno failure.