Is a test that only fails sometimes better or worse than no test?


A test that fails only sometimes (and passes sometimes) does not tell you anything more than you know without that test: there might be a bug. You either need to refine your test to isolate the bug or remove it so that you can continue working.


(IMHO) As with many other things: it depends :slight_smile: If the test only needs some fixing but it’s more or less correct, than it is better than no tests to begin with. At least it points you to the right direction. If, instead, the test is completely wrong it’s worse than useless.