I want to wait for some 'ok'-message. However, while
I'm waiting, an error dialog may appear, while the
'ok'-message will never appear. Just to wait for a
timeout isn't really helpful, as the test will be part
of a suite. Only the test should fail, not the whole
suite, and it should appear as soon as the error dialog
pops up. Thus, testing for both cases should be
supported, one will fail the test, the other not.