Consider transformed container's scroll offset when calculating fixed descendant's offset

Consider transformed container's scroll offset when calculating fixed descendant's offset

Fixed-position descendant of a transformed-container behaves differently
than normal fixed-position elements. See
http://dev.w3.org/csswg/css-transforms/#transform-rendering for details.

Though the topic is still controversial
https://www.w3.org/Bugs/Public/show_bug.cgi?id=16328 we have already
followed the current spec in most parts of our code.

The spec doesn't say if the fixed-pos element should scroll if the
transformed container is scrollable. We are inconsistent about this.
We consider the scroll offset during painting, but not during
invalidation. When the fixed-position elements needs repaint,
we invalidate the place that is different from where paint the fixed-pos
element, if the transformed container is scrolled.

This change make it consistent.

With this change, we'll avoid corrupted rendering, and behave the same
as Firefox.

BUG=439930
TEST=fast/repaint/fixed-descendant-of-transformed-scrolled.html

Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=188245

[Xianzhu, 2015-01-08 18:57:07]

wangxianzhu@chromium.org changed reviewers:
+ jchaffraix@chromium.org, leviw@chromium.org

[Julien - ping for review, 2015-01-09 13:25:01]

jchaffraix@chromium.org changed reviewers:
+ dsinclair@chromium.org

[Julien - ping for review, 2015-01-09 13:25:02]

https://codereview.chromium.org/839103002/diff/1/Source/core/rendering/Render...
File Source/core/rendering/RenderLayer.cpp (right):

https://codereview.chromium.org/839103002/diff/1/Source/core/rendering/Render...
Source/core/rendering/RenderLayer.cpp:1388: if (RenderLayerScrollableArea*
scrollableArea = ancestorLayer->scrollableArea())
Won't this add the RenderView's scrollOffset to the location as the check is
only if fixedPositionContainerLayer == ancestorLayer (not if you have
encountered a containing block for fixed positioned elements that is not the
RenderView).

[Xianzhu, 2015-01-09 17:54:16]

https://codereview.chromium.org/839103002/diff/1/Source/core/rendering/Render...
File Source/core/rendering/RenderLayer.cpp (right):

https://codereview.chromium.org/839103002/diff/1/Source/core/rendering/Render...
Source/core/rendering/RenderLayer.cpp:1388: if (RenderLayerScrollableArea*
scrollableArea = ancestorLayer->scrollableArea())
On 2015/01/09 13:25:02, Julien Chaffraix - CET wrote:
> Won't this add the RenderView's scrollOffset to the location as the check is
> only if fixedPositionContainerLayer == ancestorLayer (not if you have
> encountered a containing block for fixed positioned elements that is not the
> RenderView).

The RenderView case has been handled above (line 1340 to 1351). Added an assert
and comments here. The distance to that context too big to be easily discovered
by readers.

[Xianzhu, 2015-01-12 17:35:21]

On 2015/01/09 17:54:16, Xianzhu wrote:
>
https://codereview.chromium.org/839103002/diff/1/Source/core/rendering/Render...
> File Source/core/rendering/RenderLayer.cpp (right):
> 
>
https://codereview.chromium.org/839103002/diff/1/Source/core/rendering/Render...
> Source/core/rendering/RenderLayer.cpp:1388: if (RenderLayerScrollableArea*
> scrollableArea = ancestorLayer->scrollableArea())
> On 2015/01/09 13:25:02, Julien Chaffraix - CET wrote:
> > Won't this add the RenderView's scrollOffset to the location as the check is
> > only if fixedPositionContainerLayer == ancestorLayer (not if you have
> > encountered a containing block for fixed positioned elements that is not the
> > RenderView).
> 
> The RenderView case has been handled above (line 1340 to 1351). Added an
assert
> and comments here. The distance to that context too big to be easily
discovered
> by readers.

[Xianzhu, 2015-01-12 17:35:46]

ping...

[dsinclair, 2015-01-12 19:28:56]

The CQ bit was checked by dsinclair@chromium.org

[dsinclair, 2015-01-12 19:28:56]

lgtm

[commit-bot: I haz the power, 2015-01-12 19:29:55]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/839103002/20001

[Xianzhu, 2015-01-12 19:31:19]

Julien, I'm landing this patch as I think your comment has been addressed. The
bug is M-41 ReleaseBlock-Beta so I'd like to catch up with the first beta
release.

[commit-bot: I haz the power, 2015-01-12 20:38:53]

Committed patchset #2 (id:20001) as
https://src.chromium.org/viewvc/blink?view=rev&revision=188245