Community
Participate
Working Groups
Build Identifier: I basically reduced all the timeout,sleep,etc. values to have the test running faster in the proposed patch. I don't expect any issues with it, but it might cause timing issues on certain platforms. So we need to keep this in mind. I also tried to use and share a single httpClient. But it didn't work with one of the tests and the speed gain was not so big. So I didn't invest more time in this. Server is shared now though. Reproducible: Always
Created attachment 200290 [details] proposed patch
Runtime reduced from >10s to ~2s
Thomas, I've committed your patch with a few changes intended to make the test more readable. I also tested your changes on a slow machine, and fixed one of the test cases that failed in that environment. The end result run time is about 2.5s. Cheers, Michael
remember when addressing these long test issues, it is often best to go to a solution that uses latches rather than just reducing timeouts. Timeouts will always be non-deterministic and latches can often reduce a 10s test to milliseconds as their is no need to wait more than minimal time.
Resolved -> Closed
