We have a lot of test failures. Most are innocent: due log files made obsoletee by code changes.
This kind of thing should be eliminated: it makes using the tests much harder for everyone else. It is understandable that this happens once in a while. But these divergences must be fixed ASAP. We seem to let them drag on for a long time. Ideally, one should check in test log updates with the code change that causes them. Of course one should be careful: we don't want to bless a bug. Part of making a change is being sure that the new log is correct. Before checking it in. I too often check in changes without testing. That's because testing is so unwieldy. This is a circular problem: test failures that I might have caused get lost in the noise of endemic failure. Testing matters. Let's make it great again. _______________________________________________ Swan-dev mailing list [email protected] https://lists.libreswan.org/mailman/listinfo/swan-dev
