cc: DelegatedRenderer layer needs a surface if rotated.

cc: DelegatedRenderer layer needs a surface if rotated.

When the DRLayer is rotated to a non-axis-aligned position, it needs to
own a render surface in order to clip its delegated quads correctly.

Tests:
DelegatedRendererLayerImplTestSimple.DoesOwnARenderSurfaceForTransform

BUG=171067
R=piman


Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=177858

[danakj, 2013-01-19 02:16:23]

I am going to give DRLayer maskToBounds behaviour all the time for now, because
as we discussed it would generally never look good to not. We can look into it
in the future if something else is needed.

[piman, 2013-01-19 02:39:09]

one slight concern, but otherwise LGTM

https://codereview.chromium.org/11929045/diff/1/cc/layer_tree_host_common.cc
File cc/layer_tree_host_common.cc (right):

https://codereview.chromium.org/11929045/diff/1/cc/layer_tree_host_common.cc#...
cc/layer_tree_host_common.cc:135: return layer->masksToBounds() ||
layer->maskLayer() || layer->hasDelegatedContent();
I wonder if this could have a bad side effect if we add a child to a DRL (not
that we'd do that normally in Aura, but with the browser tag, I dunno?), because
while it wants to clip the render passes, it may not necessarily want to clip
child layers.

[danakj, 2013-01-19 02:43:00]

https://codereview.chromium.org/11929045/diff/1/cc/layer_tree_host_common.cc
File cc/layer_tree_host_common.cc (right):

https://codereview.chromium.org/11929045/diff/1/cc/layer_tree_host_common.cc#...
cc/layer_tree_host_common.cc:135: return layer->masksToBounds() ||
layer->maskLayer() || layer->hasDelegatedContent();
On 2013/01/19 02:39:09, piman wrote:
> I wonder if this could have a bad side effect if we add a child to a DRL (not
> that we'd do that normally in Aura, but with the browser tag, I dunno?),
because
> while it wants to clip the render passes, it may not necessarily want to clip
> child layers.

In my mind the DRLayer should never have any children.

With browser tag, it should have the parent compositor as the renderer, so the
DRLayer would include all the quads from the renderer and any browser tags.
Otherwise it's impossible to do things like correct z order.

SG?

[piman, 2013-01-19 05:42:33]

On Fri, Jan 18, 2013 at 6:43 PM, <danakj@chromium.org> wrote:

>
> https://codereview.chromium.**org/11929045/diff/1/cc/layer_**
>
tree_host_common.cc<https://codereview.chromium.org/11929045/diff/1/cc/layer_tree_host_common.cc>
> File cc/layer_tree_host_common.cc (right):
>
> https://codereview.chromium.**org/11929045/diff/1/cc/layer_**
>
tree_host_common.cc#newcode135<https://codereview.chromium.org/11929045/diff/1/cc/layer_tree_host_common.cc#newcode135>
> cc/layer_tree_host_common.cc:**135: return layer->masksToBounds() ||
> layer->maskLayer() || layer->hasDelegatedContent();
> On 2013/01/19 02:39:09, piman wrote:
>
>> I wonder if this could have a bad side effect if we add a child to a
>>
> DRL (not
>
>> that we'd do that normally in Aura, but with the browser tag, I
>>
> dunno?), because
>
>> while it wants to clip the render passes, it may not necessarily want
>>
> to clip
>
>> child layers.
>>
>
> In my mind the DRLayer should never have any children.
>
> With browser tag, it should have the parent compositor as the renderer,
> so the DRLayer would include all the quads from the renderer and any
> browser tags. Otherwise it's impossible to do things like correct z
> order.
>
> SG?
>

LGTM for now, we'll burn that bridge when it comes to it.


>
>
https://codereview.chromium.**org/11929045/<https://codereview.chromium.org/1...
>

[commit-bot: I haz the power, 2013-01-19 05:46:19]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/danakj@chromium.org/11929045/1

[commit-bot: I haz the power, 2013-01-20 00:38:29]

Change committed as 177858