On Thu Sep 1 16:30:06 2022 +0000, Jinoh Kang wrote:
"Flaky" is effectively equivalent to "skip" except that the test itself is always run. Ideally we could have ~~testbot~~ CI detect noticeable changes in success/failure frequency over the last _N_ test runs, and flag such occurences as unexpected success/failure. Otherwise, tests marked flaky risk going stale; this applies the same for skipped tests.
skip() would imply that we know beforehand whether the test is going to fail or succeeds. But that's not the case.