I’m going to adjust the new summary output for expected timeouts. I hope to do that in the next hour or less. I’ll put that in and flip the default back on for using the new summary output.
I’ll do those two changes separately, so you can revert the flip back on to flip it back off if we still have an issue.
(This can be orthogonal to the new work to mark up expected timeouts).
I went ahead and added the expected timeout support in r255363.
I’m going to turn back on the new BasicResultsFormatter as the default. We can flip this back off if it is still not doing everything we need, but I think we cover the issue you saw now.
Hey Pavel and/or Tamas,
Let me know when we’re definitely all clear on the expected timeout support I added to the (now once again) newer default test results.
As soon as we don’t need the legacy summary results anymore, I’m going to strip out the code that manages it. It is quite messy and duplicates the content that is better handled by the test event system.
we've had an unrelated breaking change, so the buildbots were red over
the weekend. I've fixed it now, and it seems to be turning green.
We've also had power outage during the weekend and not all of the
buildbots are back up yet, as we need to wait for MTV to wake up. I'd
like to give this at least one more day, to give them a chance to
stabilize. Is this blocking you from making further changes to the
test event system?
Oh yeah, that’s fine. I won’t take that code out.
Hmm at least some of the builds went through this weekend, I made a number of changes Saturday morning (US Pacific time) that I saw go through the Ubuntu 14.04 cmake bot.
I have removed all of our expected timeouts from dosep.py (there are
still some freebsd and darwin ones left, but I don't know If anyone is
looking at those), so I think we're not using any part of the old test
runner at the moment. All clear for removal on our part.
Great. Actually there is a latent bug in there we hit yesterday, when there is a platform set but it doesn’t start with “remote-” (we assume the match result has a first matched group and blow up). I’ll submit a patch that both fixes that up and strips out the darwin ones. I think the rerun logic now fully covers us on the Darwin side.