[SVG] Poor filter resolution when using CSS 3D transforms.

[SVG] Poor filter resolution when using CSS 3D transforms.

calculateTransformationToOutermostCoordinateSystem() should stop at the
first composited layer to match its backing resolution. Currently, we
are incorrectly taking the layer transform into account (the stop
condition is off by one).

R=pdr@chromium.org,schenney@chromium.org,senorblanco@chromium.org

BUG=345441
Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=169472

[Stephen White, 2014-03-18 17:31:33]

LGTM

[Stephen Chennney, 2014-03-18 17:59:52]

LGTM. I suppose the call to compositingState() must happen during paint, so we
do not hit the chicken-and-egg assertions in Debug.

[f(malita), 2014-03-18 18:32:51]

On 2014/03/18 17:59:52, Stephen Chenney wrote:
> LGTM. I suppose the call to compositingState() must happen during paint, so we
> do not hit the chicken-and-egg assertions in Debug.

Or not - looks like it's causing a bunch of debug crashes...

RenderSVGInlineText::computeNewScaledFontForStyle() must be calling this during
layout :(

That explains why we were using hasCompositedLayerMapping() in the first place,
in spite of comments suggesting it's not the right way to check for compositing.
I guess we can leave the compositing check alone for now, it's unrelated to this
fix.

[f(malita), 2014-03-18 18:56:56]

On 2014/03/18 18:32:51, Florin Malita wrote:
> On 2014/03/18 17:59:52, Stephen Chenney wrote:
> > LGTM. I suppose the call to compositingState() must happen during paint, so
we
> > do not hit the chicken-and-egg assertions in Debug.
> 
> Or not - looks like it's causing a bunch of debug crashes...
> 
> RenderSVGInlineText::computeNewScaledFontForStyle() must be calling this
during
> layout :(
> 
> That explains why we were using hasCompositedLayerMapping() in the first
place,
> in spite of comments suggesting it's not the right way to check for
compositing.
> I guess we can leave the compositing check alone for now, it's unrelated to
this
> fix.

On second thought, it's probably a bad idea to sneak around compositingState()
and use a possibly misleading hasCompositedLayerMapping().

I think we need to give up traversing the layer tree when
layer->isAllowedToQueryCompositingState() == false.

Updated.

[f(malita), 2014-03-18 19:35:10]

The CQ bit was checked by fmalita@chromium.org

[commit-bot: I haz the power, 2014-03-18 19:35:28]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/fmalita@chromium.org/196413023/60001

[commit-bot: I haz the power, 2014-03-18 20:42:45]

Change committed as 169472