condition-based-waiting

Replace fixed delays like setTimeout or sleep with condition polling. Use a waitFor function that checks for actual state changes at regular intervals instead of guessing how long an operation takes, eliminating race conditions that cause tests to pass inconsistently.

Wait for real conditions instead of arbitrary delays. Implement polling-based helpers like waitForEvent or waitForEventCount that verify the expected state has actually occurred, rather than sleeping and hoping the system is ready.

Yes. Condition polling handles flaky tests under load and parallel execution by eliminating timing assumptions. Tests become reliable regardless of system speed or concurrent operations, since they wait for proven state changes, not elapsed time.

Use fixed timeouts only when testing actual timing behavior itself. For all other async scenarios—waiting for events, state changes, or API responses—condition polling is more reliable and maintainable than arbitrary sleep durations.

Polling checks for conditions at 10ms intervals until they're met or a timeout expires. This approach catches state changes immediately rather than gambling on a fixed delay, preventing tests from failing on slow machines or under resource contention.