On Mon Jan 27 16:51:37 2025 +0000, William Horvath wrote:
In that case, what if I just lower the iteration count for any test case to 20, and don't break early at all? Then it simplifies the test to just making sure we don't see STATUS_TIMEOUT.
One iteration (that is, no iterations) is enough to see it. It is still flaky if you wait for yield to happen or not during given number of iterations. Making it this way with iterations doesn't make test more clear, but I'd personally wouldn't care too much if that wasn't flaky additionally.