py.test has this curious feature 'xfail', that allows you to ensure a test execute, but expect it to fail. Not sure I've had a need for it.