Bug 86886 – [chromium] Don't force the visibleLayerRect to be empty for animating layers whose front face is not visible

• Component: New Bugs
• Status: RESOLVED FIXED

[Dana Jansens]

[chromium] Don't force the visibleLayerRect to be empty for animating layers whose front face is not visible

[Dana Jansens]

Some context discussion on https://bugs.webkit.org/show_bug.cgi?id=82571

[Dana Jansens]

Created attachment 142770 [details]
Patch

[Adrienne Walker]

Comment on attachment 142770 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=142770&action=review

Hmm.  This assumes that the current visible rect (flipped) is a good proxy for the final visible rect when the animation is done.  There are certainly edge cases where that's not true, but I guess this is just as much as heuristic as "paint everything only if it's a small layer".  I think overall that this is an improvement, and I'm happy to get rid of the empty IntRect special case.

> Source/WebKit/chromium/tests/CCLayerTreeHostCommonTest.cpp:-1325
> -
> -    // But if the back face is visible, then the visibleLayerRect should be empty.
> -    EXPECT_TRUE(animatingChild->visibleLayerRect().isEmpty());
> -    EXPECT_TRUE(animatingSurface->visibleLayerRect().isEmpty());
> -    // And any layers in the subtree should not be considered visible either.
> -    EXPECT_TRUE(childOfAnimatingSurface->visibleLayerRect().isEmpty());

Can you verify that the visible layer rect for these animating layers is the whole layer (assuming it's not clipped)?

[Dana Jansens]

Comment on attachment 142770 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=142770&action=review

>> Source/WebKit/chromium/tests/CCLayerTreeHostCommonTest.cpp:-1325
>> -    EXPECT_TRUE(childOfAnimatingSurface->visibleLayerRect().isEmpty());
> 
> Can you verify that the visible layer rect for these animating layers is the whole layer (assuming it's not clipped)?

yup!

[Dana Jansens]

Created attachment 143376 [details]
Patch

I removed the change in the prepainting code. While the flipped case we should now have a non-empty rect, for animating layers that are not flipped but transitioning into the viewport from outside, we should still prepaint them when they are small.

[Adrienne Walker]

Comment on attachment 143376 [details]
Patch

R=me.  Leaving the prepainting code makes sense.

[WebKit Review Bot]

Comment on attachment 143376 [details]
Patch

Clearing flags on attachment: 143376

Committed r118090: <http://trac.webkit.org/changeset/118090>

[WebKit Review Bot]

All reviewed patches have been landed.  Closing bug.