[webkit-reviews] review requested: [Bug 90092] [chromium] ContentLayerPainter should generate opaque rects in content space : [Attachment 150065] Patch

[bugzilla-daemon at webkit.org]

vollick at chromium.org has asked	for review:
Bug 90092: [chromium] ContentLayerPainter should generate opaque rects in
content space
https://bugs.webkit.org/show_bug.cgi?id=90092

Attachment 150065: Patch
https://bugs.webkit.org/attachment.cgi?id=150065&action=review

------- Additional Comments from vollick at chromium.org
The last patch had made it so that WebKit::WebContentLayerClient took a rect
in layer space, but returned it in canvas space. This was all to get around a
rounding error that can happen with non-integer scale factors, but it doesn't
make sense to have the rects in different spaces; it puts too much of a burden
on consumers to take pains to deal with the inconsistency. So instead, I've
left
OpaqueRectTrackingContentLayerDelegate as it was. It continues to take and
produce layer space rects, and the resulting opaque rect is transformed back
into content space in CanvasLayerTextureUpdater. This can still lead to
rounding
errors due to repeatly scaling and converting back to int rects. Eventually,
the WebKit::WebContentLayerClient interface should be modified to return opaque
rects as FloatRects so that they can be scaled a whole lot without accumulating
error before finally being converted to the content space opaque rect.