Bug 36191
Summary: | [Qt] GraphicsLayer: fix test regressions | ||
---|---|---|---|
Product: | WebKit | Reporter: | Noam Rosenthal <noam> |
Component: | Layout and Rendering | Assignee: | Noam Rosenthal <noam> |
Status: | CLOSED FIXED | ||
Severity: | Normal | CC: | hausmann, jesus, ossy |
Priority: | P3 | Keywords: | Qt |
Version: | 528+ (Nightly build) | ||
Hardware: | PC | ||
OS: | All | ||
Bug Depends on: | 36216, 36219 | ||
Bug Blocks: | 35784 |
Noam Rosenthal
Some regressions were found in test animations.
We need to check if those are actual regressions or a test issue, and fix.
Attachments | ||
---|---|---|
Add attachment proposed patch, testcase, etc. |
Noam Rosenthal
http://build.webkit.org/results/Qt%20Linux%20Release/r55955%20(8504)/results.html
Jesus Sanchez-Palencia
Is this Mac OS specific?
If not, could you please change the OS and assign this to webkit-qt-unassigned@trolltech.com so we don't add another item to the untriaged bugs list ?
(I can't edit the bugzilla entry, otherwise I would edit this.)
Noam Rosenthal
Done.
Noam Rosenthal
When running http://trac.webkit.org/export/55955/trunk/LayoutTests/animations/simultaneous-start-left.html it passes.
The other two bugs also fail on Safari windows, and are related to new and undocumented css properties.
Deferring to P3 for now.
Csaba Osztrogonác
Failing tests skipped by http://trac.webkit.org/changeset/56106 ,
please unskip them with fix.
Noam Rosenthal
For now we'll leave them as skipped - which is the right way to treat tests for new webkit features that we don't support yet in Qt :)
For some of these tests it's not a regression since they were only added a couple of weeks ago.
Noam Rosenthal
36216 is a brand new CSS feature - no point in making support for it a blocker.
Csaba Osztrogonác
I don't understand exactly why they aren't regressions. I only see these tests passed before enabling accelerated compositing by default, and failed after the patch landed. What is it if not a regression?
Noam Rosenthal
To be more specific:
* animations/fill-mode-transform.html
This is a new feature (added mid March) that we need to support in Qt AC. Of course a new feature that Apple adds that requires AC support is going to fail with Qt, because we simply haven't implemented support for that feature yet. Classic case for temporary skip and definitely not a "regression".
* animations/play-state.html
Fixed, see 36219
animations/simultaneous-start-left.html
I couldn't reproduce it - waiting till we have proper QGraphicsWebView support in DRT to look into it again.
In general we're not running tests on QGraphicsWebView regularly - until we do that there's no point in treating AC in a special way. Once all the DRT tests work on QGraphicsWebView, I can look at the delta with AC.
Noam Rosenthal
A fix is now coming.
Noam Rosenthal
all underlying bugs have been fixed.