[webkit-reviews] review requested: [Bug 106142] Promote composited-scrolling layers to stacking contexts. : [Attachment 182623] Patch

[bugzilla-daemon at webkit.org]

vollick at chromium.org has asked	for review:
Bug 106142: Promote composited-scrolling layers to stacking contexts.
https://bugs.webkit.org/show_bug.cgi?id=106142

Attachment 182623: Patch
https://bugs.webkit.org/attachment.cgi?id=182623&action=review

------- Additional Comments from vollick at chromium.org
(In reply to comment #10)
> (From update of attachment 182491 [details])
> Ugh, so much additional complexity. This would be so much simpler if you
could just whack style->zIndex() when you know that you can safely propagate
this overflow element into a stacking context. What's the downside of doing
this based simply on looking at style, not whether it actually overlows now?

I was hoping to do this, too, but the problem is that I use the value of
isStackingContext when computing m_needsCompositedScrolling. So when computing
m_needsCompositedScrolling, rather than asking if a layer is a stacking context
I really need to ask "would you be a stacking context irrespective of your
composited scrolling status?" If I promote to a stacking context by mutating
the style, I'm no longer able to ask this question because the
composited-scrolling effect is baked in.

Also in this patch:
 - Added the missing mac test expectation.
 - Updated the chromium TestExpectations for the test that will need
rebaselining.
 - Removed the optimization in RenderLayer::updateCompositedLayersAfterScroll.
Should be in its own patch anyhow.