Implement FullScreen using top layer.

This CL is based off of https://codereview.chromium.org/788073004 by jchaffraix@. 

  Replace RenderFullscreen with top layer - Take II

  This patch is strongly inspired by
  https://codereview.chromium.org/134753003/
  but is a complete rewrite of the change to:
   - understand why every bits was needed (some were not).
   - better document the differences we have with the
     specification so that we can track and fix them.

  The issue with re-attaching plugins went away due to
  https://codereview.chromium.org/23618022, which moved
  the plugin's lifecycle to the DOM. 

  The tests' changes are because fullscreen doesn't center
  by default anymore but stretches the element to the
  viewport's size. Some tests were also not accounting
  for the backdrop and where thus modified.

That change didn't land due to plugin persistence issues. Those issues should now be resolved and plugins should survive over reattach.

BUG=240576, 246077, 398599

[dsinclair, 2015-09-14 14:59:11]

Patchset #4 (id:60001) has been deleted

[dsinclair, 2015-09-14 14:59:15]

Patchset #3 (id:40001) has been deleted

[dsinclair, 2015-09-14 14:59:22]

Patchset #2 (id:20001) has been deleted

[dsinclair, 2015-09-17 18:48:24]

dsinclair@chromium.org changed reviewers:
+ esprehn@chromium.org, jchaffraix@chromium.org, leviw@chromium.org

[dsinclair, 2015-09-17 18:48:25]

PTAL.

This is the work jchaffraix@ did to port fullscreen top layer. The one
outstanding layout test I'm not sure about (you can see it from the mac test run
before it timed out). The colour of the backdrop is black, previously we would
have done it as white. I updated the test example to make it white but with
content_shell it seems to be adding scrollbars. The test looks correct when I
run locally.

So, a) is the addition of scrollbars fine and I just need to rebaseline and b)
is the addition of body::backdrop {} to the test valid or is this an actual
issue?

[dsinclair, 2015-09-21 14:15:59]

ping

[leviw_travelin_and_unemployed, 2015-09-21 20:26:22]

Few comments. Looks like you'll want to give this another go at the try jobs, as
well.

https://codereview.chromium.org/1139033006/diff/180001/Source/core/css/resolv...
File Source/core/css/resolver/StyleAdjuster.cpp (right):

https://codereview.chromium.org/1139033006/diff/180001/Source/core/css/resolv...
Source/core/css/resolver/StyleAdjuster.cpp:188: // that as we don't expose this
information. FIXME: We should find a way to
Fixme is the old. Use the new hotness: TODO(dsinclair)

Also, I'd rather see something like "TODO(dsinclair): The ability to do this
should be standardized crbug.com/___"

https://codereview.chromium.org/1139033006/diff/180001/Source/core/css/resolv...
File Source/core/css/resolver/StyleResolver.cpp (right):

https://codereview.chromium.org/1139033006/diff/180001/Source/core/css/resolv...
Source/core/css/resolver/StyleResolver.cpp:677: if (pseudoId != BACKDROP &&
!canHaveGeneratedChildren(*parentLayoutObject))
There was no way to get into this case before with dialogue? If there was, can
we do this in a separate change?

https://codereview.chromium.org/1139033006/diff/180001/Source/core/dom/Fullsc...
File Source/core/dom/Fullscreen.cpp (right):

https://codereview.chromium.org/1139033006/diff/180001/Source/core/dom/Fullsc...
Source/core/dom/Fullscreen.cpp:431: // FIXME: Why can we get here without
calling pushFullscreenElementStack?
Nit: FIXME again.

Also, it'd be good to find the answer to this question...

https://codereview.chromium.org/1139033006/diff/180001/Source/core/dom/StyleC...
File Source/core/dom/StyleChangeReason.h (right):

https://codereview.chromium.org/1139033006/diff/180001/Source/core/dom/StyleC...
Source/core/dom/StyleChangeReason.h:29: CORE_EXPORT extern const char
FullScreen[];
Why is this necessary for FullScreen but nowhere else?

[rune, 2015-09-22 11:47:56]

rune@opera.com changed reviewers:
+ rune@opera.com

[rune, 2015-09-22 11:47:57]

https://codereview.chromium.org/1139033006/diff/180001/Source/core/dom/StyleC...
File Source/core/dom/StyleChangeReason.h (right):

https://codereview.chromium.org/1139033006/diff/180001/Source/core/dom/StyleC...
Source/core/dom/StyleChangeReason.h:29: CORE_EXPORT extern const char
FullScreen[];
On 2015/09/21 20:26:22, leviw (UTCplus1 until 9-17) wrote:
> Why is this necessary for FullScreen but nowhere else?

I suspect that's because it's passed to setNeedsStyleRecalc in web/.

https://codereview.chromium.org/1139033006/diff/180001/Source/web/FullscreenC...
File Source/web/FullscreenController.cpp (right):

https://codereview.chromium.org/1139033006/diff/180001/Source/web/FullscreenC...
Source/web/FullscreenController.cpp:184:
document->setNeedsStyleRecalc(SubtreeStyleChange,
StyleChangeReasonForTracing::create(StyleChangeReason::FullScreen));
So, what is actually requiring a style recalc here?

It used to just recalculate the style for just the LayoutFullScreen, and since
it sets the width and height in px based on the viewport size, I understand why
it's necessary when changing the viewport size, but for all the other elements?

Couldn't you just use fullscreenElement->setNeedsStyleRecalc(LocalStyleChange,
...), to trigger the recalc + styleAdjustment (where the viewport px sizes are
now set)?

Also, if you implemented a method inside blink (Fullscreen for instance) that
looked up and did setNeeds..., you'd avoid the CORE_EXPORT on the FullScreen
constant. The setNeeds... calls for entering and exiting fullscreen are
contained in the Fullscreen class.

[dsinclair, 2015-09-23 19:47:59]

On 2015/09/22 11:47:57, rune wrote:
>
https://codereview.chromium.org/1139033006/diff/180001/Source/core/dom/StyleC...
> File Source/core/dom/StyleChangeReason.h (right):
> 
>
https://codereview.chromium.org/1139033006/diff/180001/Source/core/dom/StyleC...
> Source/core/dom/StyleChangeReason.h:29: CORE_EXPORT extern const char
> FullScreen[];
> On 2015/09/21 20:26:22, leviw (UTCplus1 until 9-17) wrote:
> > Why is this necessary for FullScreen but nowhere else?
> 
> I suspect that's because it's passed to setNeedsStyleRecalc in web/.
> 
>
https://codereview.chromium.org/1139033006/diff/180001/Source/web/FullscreenC...
> File Source/web/FullscreenController.cpp (right):
> 
>
https://codereview.chromium.org/1139033006/diff/180001/Source/web/FullscreenC...
> Source/web/FullscreenController.cpp:184:
> document->setNeedsStyleRecalc(SubtreeStyleChange,
> StyleChangeReasonForTracing::create(StyleChangeReason::FullScreen));
> So, what is actually requiring a style recalc here?
> 
> It used to just recalculate the style for just the LayoutFullScreen, and since
> it sets the width and height in px based on the viewport size, I understand
why
> it's necessary when changing the viewport size, but for all the other
elements?
> 
> Couldn't you just use fullscreenElement->setNeedsStyleRecalc(LocalStyleChange,
> ...), to trigger the recalc + styleAdjustment (where the viewport px sizes are
> now set)?
> 
> Also, if you implemented a method inside blink (Fullscreen for instance) that
> looked up and did setNeeds..., you'd avoid the CORE_EXPORT on the FullScreen
> constant. The setNeeds... calls for entering and exiting fullscreen are
> contained in the Fullscreen class.


Continuing here: https://codereview.chromium.org/1363023005