Flood-fill the canvas used for layout tests

Flood-fill the canvas used for layout tests

Otherwise, layout tests that don't damage the entire view access
uninitialized memory.

BUG=332271
R=reed@google.com,thestig@chromium.org

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=243694

[reed1, 2014-01-08 15:49:17]

https://codereview.chromium.org/128443002/diff/1/content/shell/renderer/test_...
File content/shell/renderer/test_runner/WebTestProxy.cpp (right):

https://codereview.chromium.org/128443002/diff/1/content/shell/renderer/test_...
content/shell/renderer/test_runner/WebTestProxy.cpp:635:
m_canvas.reset(skia::CreateBitmapCanvas(scaledWidth, scaledHeight, true));
I think you want to just pass 'false' to CreateBitmapCanvas. Otherwise, you
claiming to skia that all the pixels will be opaque, which tells skia to ignore
reading the alpha values during compositing.

By passing false, skia will flood-fill the pxiels with zero for you (and
potentially faster than you can, when we start using calloc. WIth 'false', you
won't need to call drawARGB().

[jochen (gone - plz use gerrit), 2014-01-08 18:03:51]

https://codereview.chromium.org/128443002/diff/1/content/shell/renderer/test_...
File content/shell/renderer/test_runner/WebTestProxy.cpp (right):

https://codereview.chromium.org/128443002/diff/1/content/shell/renderer/test_...
content/shell/renderer/test_runner/WebTestProxy.cpp:635:
m_canvas.reset(skia::CreateBitmapCanvas(scaledWidth, scaledHeight, true));
On 2014/01/08 15:49:17, reed1 wrote:
> I think you want to just pass 'false' to CreateBitmapCanvas. Otherwise, you
> claiming to skia that all the pixels will be opaque, which tells skia to
ignore
> reading the alpha values during compositing.
> 
> By passing false, skia will flood-fill the pxiels with zero for you (and
> potentially faster than you can, when we start using calloc. WIth 'false', you
> won't need to call drawARGB().

Done.

[reed1, 2014-01-08 18:13:06]

lgtm

[reed1, 2014-01-08 18:13:56]

On 2014/01/08 18:13:06, reed1 wrote:
> lgtm

I wonder if a comment might be in order, to document why you're allocating a
buffer to be non-opaque (the false) even thought you expect the content itself
to be opaque...

[jochen (gone - plz use gerrit), 2014-01-08 19:29:48]

On 2014/01/08 18:13:56, reed1 wrote:
> On 2014/01/08 18:13:06, reed1 wrote:
> > lgtm
> 
> I wonder if a comment might be in order, to document why you're allocating a
> buffer to be non-opaque (the false) even thought you expect the content itself
> to be opaque...

done

[commit-bot: I haz the power, 2014-01-08 19:30:49]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/jochen@chromium.org/128443002/150001

[commit-bot: I haz the power, 2014-01-08 23:32:42]

Change committed as 243694

[alph, 2014-01-09 06:56:13]

A revert of this CL has been created in
https://codereview.chromium.org/130963002/ by alph@chromium.org.

The reason for reverting is: a bunch of video* layout tests failed
http://build.chromium.org/p/chromium.webkit/builders/WebKit%20Linux/builds/24977.