[Webkit-unassigned] [Bug 86796] Change a bunch of things in the test_expectations.txt format

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=86796





--- Comment #12 from Dirk Pranke <dpranke at chromium.org>  2012-05-18 11:36:41 PST ---
(In reply to comment #11)
> (In reply to comment #10)
> > (In reply to comment #9)
> > > > Hm. I'm strongly tempted to kill SLOW instead of implement this :).
> > > 
> > > Please don't mix syntax changes with functionality changes in the same patch. If we want to change the meaning of anything, that should be in an isolated patch.
> > 
> > Of course. I wouldn't be likely to make all of these changes in one patch, either. Dunno, I am torn on SLOW ...
> 
> If we got rid of slow, we'd need to change a few other things to compensate.
> 1. We'd need to increase the timeout considerably.

Yeah, I was thinking we'd match the default timeout the other ports use (30s)

> 2. We'd need to decrease the timeout for tests expected to timeout in order to maintain buildbot cycle times.
>

I could see how this would be confusing. I'd be inclined to not do this.

I am wondering if there's really a point to running tests that TIMEOUT at all (or CRASH, for that matter). Maybe they should be skipped by default. The downside would be that you wouldn't be able to tell if a test started passing, and that's a pretty big downside.

Another idea I have considered is breaking the steps on the bot into a step that runs just the stable tests (no flaky tests, crashes, or timeouts), and have a second step or a second bot that runs the others. I don't think this is a big enough win to justify the complexity, though.

We should probably break the SLOW discussion off into an entirely different bug.

-- 
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.