[Webkit-unassigned] [Bug 161242] [GTK][Threaded Compositor] Several flaky tests

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=161242

--- Comment #16 from Carlos Garcia Campos <cgarcia at igalia.com> ---
(In reply to comment #15)
> (In reply to comment #14)
> > Finally found the issue. The problem is the viewport controller, that is
> > changed by the fixed layout tests, but never reset. There's also a problem
> > with the fixed layout tests that could leave the view in an inconsistent
> > state as I explained in bug #160117. I'm already working on a proper fix,
> > but in the meantime I've skipped fixed layout tests (they are only two) to
> > make sure we don't see all those flaky tests in the bots anymore. To easily
> > reproduce it, launch run-webkit-tests with a single worker job and run fixed
> > layout test first and then other ref tests, for example:
> > 
> > $ run-webkit-tests --gtk --release --child-processes=1
> > fast/fixed-layout/fixed-layout.html fast/regions
> 
> It's clearly a big improvement, good job.
> 
> But it's still not completely fixed. We have 50 unexpected passes in build
> #18009, for example. These are tests that normally always fail, but
> occasionally pass now that we've enabled threaded compositor.

You are assuming again that there's a single issue. This particular problem that was making both the expected and actual rendering of reftest fail, was caused by the fixed layout (it changes the viewport controller state that is never reset). Of course there are more issues, but I think the situation now is mostly the same to what we had before switching to the threaded compositor.

-- 
You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.webkit.org/pipermail/webkit-unassigned/attachments/20160902/c078e8a7/attachment-0001.html>