Blend background with existing content

Add setting to blend background with existing content

If both base background color (WebView::setBaseBackgroundColor) and
background color (set by document) has alpha, then the background color
should be blended into existing content.

This behavior is important in Android WebView, which is a view inside the
Android view tree and needs to blend with views below it.

Internal bug: 10750978

BUG=

PS10 committed in r163657 and reverted in r163699 due to test failing
on android since android has different color ordering.

Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=163717

[boliu, 2013-09-17 18:19:36]

Up for initial review.

No tests yet. I only implemented what I _thought_ was correct for android
webview, and have no idea if I broke anything else...any advice on how to test
this, and make sure nothing else broke?

[aelias_OOO_until_Jul13, 2013-09-17 18:25:59]

Keep an eye on layout tests in linux_blink_rel, they cover rendering correctness
issues.

[boliu, 2013-09-17 23:36:05]

Err, PS3 broke fast/css/empty-webkit-mask-crash.html

[boliu, 2013-09-24 16:43:58]

+enne for initial review

Enne, do you think this change is reasonable? If so, I can add proper layout
tests for this.

[enne (OOO), 2013-09-24 18:35:46]

This approach seems reasonable to me.

https://codereview.chromium.org/23483051/diff/27001/Source/core/rendering/Ren...
File Source/core/rendering/RenderBoxModelObject.cpp (right):

https://codereview.chromium.org/23483051/diff/27001/Source/core/rendering/Ren...
Source/core/rendering/RenderBoxModelObject.cpp:716: bool
shouldBlendWithExistingContent = document().settings() &&
document().settings()->shouldBlendWithExistingContent();
bikeshed: shouldBlendWithExistingContent is a little bit ambiguous in the
context of a document setting.  (What's being blended? What content?) Maybe
something like shouldClearDocumentBackground? Or, maybe there's some other
better name?

[enne (OOO), 2013-10-09 16:56:36]

Is there a way to test this?

[boliu, 2013-10-09 18:30:23]

On 2013/10/09 16:56:36, enne wrote:
> Is there a way to test this?

My current thought is need to put a fixed pattern into the skcanvas before
normal blink drawing starts. Then do pixel test that the pattern shows up on the
final page.

[boliu, 2013-12-10 02:26:20]

Finally back to upstreaming this android webview change. This did ship in
kitkat.

Essentially no change since 3 months ago. Added a WebViewTest.

[aelias_OOO_until_Jul13, 2013-12-10 04:02:56]

Source/web lgtm, over to enne@ for Source/core.

[enne (OOO), 2013-12-10 18:25:42]

lgtm2

[boliu, 2013-12-10 18:37:04]

+jamesr for public

[jamesr, 2013-12-10 23:40:47]

public/ lgtm

[commit-bot: I haz the power, 2013-12-11 00:00:09]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/boliu@chromium.org/23483051/57001

[commit-bot: I haz the power, 2013-12-11 01:08:37]

Change committed as 163657

[johnme, 2013-12-11 11:26:14]

A revert of this CL has been created in
https://codereview.chromium.org/110913005/ by johnme@chromium.org.

The reason for reverting is:
WebViewTest.SetBaseBackgroundColorAndBlendWithExistingContent is failing
consistently on WebKit Android (Nexus4) and Android Tests (dbg) as
WebCore::redChannel(color) evaluates to false.

[aelias_OOO_until_Jul13, 2013-12-11 16:32:09]

Looks like the reason is that you looked directly at the byte data of the bitmap
in your test.  But the Skia internal color ordering is different on Android. 
Use SkBitmap::getColor(x, y) instead.  Sorry for not catching this during code
review.

[boliu, 2013-12-11 16:33:00]

On 2013/12/11 16:32:09, aelias wrote:
> Looks like the reason is that you looked directly at the byte data of the
bitmap
> in your test.  But the Skia internal color ordering is different on Android. 
> Use SkBitmap::getColor(x, y) instead.  Sorry for not catching this during code
> review.

Ahh, thanks.

[commit-bot: I haz the power, 2013-12-11 17:26:10]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/boliu@chromium.org/23483051/77001

[commit-bot: I haz the power, 2013-12-11 18:28:24]

Change committed as 163717