Sometimes one detects a condition that should be fixed, but is not severe enough to warrant failing a test altogether. One such example are warnings reported through Python's warnings module. So far, these get swallowed up by py.test. I would argue that they should be reported somehow.
Issue #108 new