On Wed, 3 May 2023 19:54:46 GMT, Phil Race <p...@openjdk.org> wrote: > The delay -> sleep changes really are no-ops .. except one you made shorter > .. And the calls like validate() seem OK .. Do you have any explanation as to > why these changes fix the time out ?
The original test had this shorter delay, so I experimented with reverting the change here. After your suggestion to make ```passed``` volatile, the test passes on windows. The original intermittent failure due to timeout seems to have been caused by the setAutoWaitForIdle() call. On my local machine, this line causes the test to take much longer to start for some reason. After removing this, the test consistently passes on 100 repeat runs of the test on all OS's except Windows. Windows fails intermittently as well, but changing the ```passed``` var to volatile fixes this. Now all OS's are passing 100 repeat runs. ------------- PR Comment: https://git.openjdk.org/jdk/pull/13761#issuecomment-1535552643
