Comment 6 for bug 935525

David Benjamin (davidben) wrote :

> That timing is not exact, and an ideal fix would be event-based rather than wallclock seconds, but doubling the time allowed is generous.

It's also worth noting that the test is testing something event-based in the first place. It's testing that child watches fire events when the child dies. The timeout is only there if it fails and hangs. So this doesn't mean the test is flakey and uses ad-hoc timeouts. It means the test is failing.