New issue
Advanced search Search tips
Note: Color blocks (like or ) mean that a user may not be available. Tooltip shows the reason.
Starred by 9 users

Issue metadata

Status: WontFix
Owner: ----
Closed: Jun 2014
EstimatedDays: ----
NextAction: ----
OS: Windows
Pri: 2
Type: Bug

issue 8606

Sign in to add a comment

Issue 364653: lots of failures running the layout tests on win x64

Reported by, Apr 17 2014 Project Member

Issue description

Hot on the heels of me actually trying to do this (see bug 178207), it looks like there are a large number of things to investigate, and very few of them were the sort of tiny pixel-diff failures I expect from a compiler or architecture change.

Running the tests (using the -t Release_x64 arg), I got ~450 failures. Many of them were in the windows-directwrite suite, but the failures there looked like cases where we were seeing characters moved by multiple pixels (and the failures were not happing on the 32-bit build at the same revision).

There were a lot of other failures as well, though. Oddly, many of the shadow dom tests were failing, as if it wasn't enabled at all. There were failures in rubberbanding, and softwarecompositing, and editing as well.

I saw no obvious indications that there was anything wrong with run-webkit-tests or content_shell that would be causing these failures, unfortunately.

Someone else should probably attempt to reproduce my results, and then we should figure out a plan to triage and deal w/ these failures.  

I unfortunately do not have much time to devote to this at the moment.

Comment 1 by, Apr 18 2014

Blocking: chromium:8606

Comment 2 by, Jun 19 2014

Status: WontFix
bah. turns out that -t Release_x64 actually does what it's suppose to, and creates a new "Release_x64" config that isn't either "Debug" or "Release"; as a result, *none* of the existing TestExpectations match it, and so nearly all of these failures were things you'd expect to fail.

Once we try using something closer to a proper "Release" configuration, you get a handful of failures that are the sort I'd expect.

I'm closing this as invalid. I will deal w/ the remaining failures and baselines as part of bug 178207. Sorry for the wasted time!

Comment 3 by, Jun 19 2014

So, somewhat amusingly, it looks like there may be a grand total of *2* legit test failures on x64, and those just need new baselines ...

Comment 4 by, May 18 2016

Labels: Test-Layout

Sign in to add a comment