[WIP] Migrate Fullscreen to use top layer instead of RenderFullScreen

[WIP] Migrate Fullscreen to use top layer instead of RenderFullScreen

BUG=240576

[falken, 2013-07-09 12:50:22]

This patch is WIP. Uploading now to share what code changes happen with the
migration.

Surprisingly only a few layout tests fail with this change (at least on my
Linux). The failing tests are:

fullscreen/full-screen-placeholder.html
(This is expected to fail, as there's is no placeholder mechanism for top
layer.)

fullscreen/full-screen-zIndex.html
(This is just due to positioning in top layer being different from the flexbox.)

fullscreen/parent-flow-inline-with-block-child.html	
(This is a progression; upon exiting fullscreen the elements are laid out as
expected now.)

plugins/fullscreen-plugins-dont-reload.html
(This is the only real failure. It tests that the element doesn't get detached
when entering fullscreen, but  necessarily happens for top layer.
http://wkbug.com/66531)

https://codereview.chromium.org/18896003/diff/1/LayoutTests/fullscreen/full-s...
File LayoutTests/fullscreen/full-screen-iframe-zIndex-expected.txt (right):

https://codereview.chromium.org/18896003/diff/1/LayoutTests/fullscreen/full-s...
LayoutTests/fullscreen/full-screen-iframe-zIndex-expected.txt:1: This tests that
an element with a positive z-index appears behind the full screen element. After
entering full screen mode, only a "Go full screen" button against a white
background should be visible. Click go full screen to run the test.
I'm not changing the expectation here. Just clarifying this description.

https://codereview.chromium.org/18896003/diff/1/Source/core/css/fullscreen.css
File Source/core/css/fullscreen.css (right):

https://codereview.chromium.org/18896003/diff/1/Source/core/css/fullscreen.cs...
Source/core/css/fullscreen.css:4: }
Ideally the above can be removed. But there is some weird behavior when making
an iframe fullscreen. Have to investigate more.

The other non-spec'd styles in this file should also be removed.

https://codereview.chromium.org/18896003/diff/1/Source/core/css/fullscreen.cs...
Source/core/css/fullscreen.css:27: 
These two are straight from the spec (:fullscreen renamed to
:-webkit-full-screen to get it to work though).

https://codereview.chromium.org/18896003/diff/1/Source/core/dom/FullscreenCon...
File Source/core/dom/FullscreenController.cpp (left):

https://codereview.chromium.org/18896003/diff/1/Source/core/dom/FullscreenCon...
Source/core/dom/FullscreenController.cpp:391:
document()->recalcStyle(Node::Force);
I think this recalcStyle was a quirk needed for RenderFullScreen. I'll hunt down
when it was added.

https://codereview.chromium.org/18896003/diff/1/Source/core/rendering/RenderV...
File Source/core/rendering/RenderVideo.cpp (left):

https://codereview.chromium.org/18896003/diff/1/Source/core/rendering/RenderV...
Source/core/rendering/RenderVideo.cpp:310: 
These were added in https://bugs.webkit.org/show_bug.cgi?id=62400

It's unclear why removing them doesn't seem to break any tests.

[falken, 2013-07-10 02:30:00]

> plugins/fullscreen-plugins-dont-reload.html
> (This is the only real failure. It tests that the element doesn't get detached
> when entering fullscreen, but  necessarily happens for top layer.
> http://wkbug.com/66531)

Yeah this must be fixed. Otherwise when you fullscreen a flash video it will
restart at the beginning
(can test this by calling requestFullScreen manually on the plugin or an
ancestor; most flash sites use flash's fullscreen feature.
thanks to jeremya@ for the info)

[trchen, 2013-07-18 22:26:29]

Hey falken,
I'm working on USE(NATIVE_FULLSCREEN_VIDEO) to make it work with subtitles. I
know this is an early WIP. Just trying to study it to make sure my code will be
compatible with your approach. :)

https://codereview.chromium.org/18896003/diff/1/Source/core/css/fullscreen.css
File Source/core/css/fullscreen.css (right):

https://codereview.chromium.org/18896003/diff/1/Source/core/css/fullscreen.cs...
Source/core/css/fullscreen.css:10: bottom: 0;
nits: Over-constrained. Right and bottom will be ignored here.

https://codereview.chromium.org/18896003/diff/1/Source/core/dom/FullscreenCon...
File Source/core/dom/FullscreenController.cpp (left):

https://codereview.chromium.org/18896003/diff/1/Source/core/dom/FullscreenCon...
Source/core/dom/FullscreenController.cpp:384: }
I'm not sure how Document::addToTopLayer() works, but wouldn't it cause the page
to be reflowed without using a placeholder box?

[falken, 2013-07-22 17:51:44]

Thanks for the comments.

https://codereview.chromium.org/18896003/diff/1/Source/core/css/fullscreen.css
File Source/core/css/fullscreen.css (right):

https://codereview.chromium.org/18896003/diff/1/Source/core/css/fullscreen.cs...
Source/core/css/fullscreen.css:10: bottom: 0;
On 2013/07/18 22:26:29, trchen wrote:
> nits: Over-constrained. Right and bottom will be ignored here.

Good point. I'll file a bug with the spec as this is copy and pasted from it.

https://codereview.chromium.org/18896003/diff/1/Source/core/dom/FullscreenCon...
File Source/core/dom/FullscreenController.cpp (left):

https://codereview.chromium.org/18896003/diff/1/Source/core/dom/FullscreenCon...
Source/core/dom/FullscreenController.cpp:384: }
On 2013/07/18 22:26:29, trchen wrote:
> I'm not sure how Document::addToTopLayer() works, but wouldn't it cause the
page
> to be reflowed without using a placeholder box?

Yes, I think so. I thought we should remove it since the spec doesn't mention a
placeholder, and it's adding more complexity by making the render tree differ
even more from the DOM tree.

[esprehn, 2013-12-02 05:03:47]

How's this going? I'd really like to see us get full screen moved to top layer.

[falken, 2013-12-02 05:22:36]

On 2013/12/02 05:03:47, esprehn wrote:
> How's this going? I'd really like to see us get full screen moved to top
layer.

Sorry for the delay. My top priority now is shipping <dialog>. But this patch
doesn't have to wait for <dialog>. If someone else would like to take this
patch/issue over, they're welcome to. Otherwise I can pick it back up after
shipping.

[falken, 2014-01-21 11:39:36]

I uploaded a new patch at https://codereview.chromium.org/139743005/