Closed choksi81 closed 10 years ago
Author: justinc Odd, this seems to happen because both threads get delayed for some longer time than usual. For example, it seems that they both will delay for 1.5-1.7 seconds and then run in an arbitrary order.
I can't increase the number of randomfloat calls because the OS will actually block in many cases then.
This is exacerbated by the fact that scheduling times seem to be chaotic in most cases.
Author: justinc Fixed in r4182. I only needed to update the test. It looks like this is a mix of scheduler / timer firing issues and general randomness. I do three runs and have a sleep to make timer behavior more predictable.
On Windows XP, it is common for the ut_repytests_randomratetest.py to fail. The root cause is unclear.