QA » openQA Project » openQA Tests

I agree that firefox should not be started from xterm, that's too far away from good customer experience. IMHO the usual way of debugging using any log output in e.g. ~/.xsession-errors* should be used.

I would close this and the PR with Rejected - the generic function is very handy to verify firefox is really stopped before we start the next firefox test. This might be a bit unlike what users do - but then again, users don't start firefoxes with fresh configs 10 times in a row either.

So if I understand this correctly, this is mostly a conflict between the authenticity of the test behavior on one side, and ease of test debugging and maintenance on the other side.

Personally, I can understand the reasoning behind both arguments, although I would tend to stand by coolo's suggestion.
Do we have any generic guidelines as to which approach should take the preference?

In doubt do as a manual test would be outlined. If the test case is 'start firefox on clean config', then I find it natural to start it from a terminal.