We have a large automatic test suite here at Net-a-Porter, and that test suite gets run automatically whenever we check code changes in to our source code repository. Of course, a developer will run the suite manually before checking in changes, but it’s nice for the rest of the team to be able to see that a particular change hasn’t broken anything.
Occasionally, the test suite will flag problems that weren’t introduced with the most recent code change. This can get frustrating for the poor developer who gets the blame for breaking something that she didn’t touch. So when we find strange problems like this, we try to fix them; this is how we fixed one this morning.