[webkit-reviews] review requested: [Bug 183040] Separate paint and scroll offsets for RenderLayerBacking::m_scrollingContentsLayer : [Attachment 334454] Patch

[bugzilla-daemon at webkit.org]

Frédéric Wang (:fredw) <fred.wang at free.fr> has asked  for review:
Bug 183040: Separate paint and scroll offsets for
RenderLayerBacking::m_scrollingContentsLayer
https://bugs.webkit.org/show_bug.cgi?id=183040

Attachment 334454: Patch

https://bugs.webkit.org/attachment.cgi?id=334454&action=review




--- Comment #2 from Frédéric Wang (:fredw) <fred.wang at free.fr> ---
Comment on attachment 334454
  --> https://bugs.webkit.org/attachment.cgi?id=334454
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=334454&action=review

@Simon, Daniel: This patch just extracts the scrollOffset from
offsetFromRender, but I'm not sure it is what was meant in the FIXME nor
whether it's really an improvement... WDYT?

> Source/WebCore/rendering/RenderLayerBacking.cpp:2471
>	   // but the repaint rect is computed without taking the scroll
position into account (see shouldApplyClipAndScrollPositionForRepaint()).

I'm a bit confused by this comment. The first line (from Daniel, bug 125239)
seems to say it is done because scrollOffset is included in the
offsetFromRenderer in RenderLayerBacking::updateGeometry(), which actually
happens on non-iOS platforms too and what I'm trying to remove here. The second
line (from Simon, bug 159186) seems to say it is done because scrollPosition in
applyCachedClipAndScrollPositionForRepaint (rather than
shouldApplyClipAndScrollPositionForRepaint). So it does not seem we can get rid
of this...