test determinism: the idle-probe test's setTimeout calls are its input, not a sleep (base red after #34155)

Merged
TE
3d ago
+1-0

Base red fix: cmux-next CI guard group โ†’ workflow-guard-tests / quality-determinism: Validate test determinism gate is red on feat-cmux-next, for example https://github.com/manaflow-ai/cmux/actions/runs/41722291773:

webviews/test/agent-pane-idle-probe.test.ts:31: sleep-then-assert: window.setTimeout(() => undefined, 5);
webviews/test/agent-pane-idle-probe.test.ts:32: sleep-then-assert: window.setTimeout(() => undefined, 5);
test-determinism: 2 active finding(s), 406 allowlisted, 408 total

#34155 (755e54b8490) added that test. These two calls don't make the test wait for state. They are what idleProbe counts: the test asserts the probe saw exactly 2 timeouts and restored setTimeout. That makes the finding a false positive, so the file gets an allowlist line with that reason. Nothing in the test is changed.

Tests: python3 scripts/check-test-determinism.py on the fleet (cmux-ci test).

Agent session: agent session

๐Ÿค– Generated with Claude Code


Autofix with [code]smith
Open in [code]smith ยท Need help on this PR? Tag @codesmith-bot with what you need.


Summary by cubic

Fixes the determinism check that went red on feat-cmux-next after #34155 added the idle-probe test. The two setTimeout calls the gate flagged aren't sleeps โ€” they're the input idleProbe counts, so the test asserts on them. Adds an allowlist entry with that reason; no test code changes.