Add UA backdrop style.

Add UA backdrop style.

This CL adds a display style to the default ::backdrop. As per 
https://fullscreen.spec.whatwg.org/#user-agent-level-style-sheet-defaults


BUG=240576

[dsinclair, 2015-09-29 13:31:22]

dsinclair@chromium.org changed reviewers:
+ esprehn@chromium.org, philipj@opera.com

[dsinclair, 2015-09-29 13:31:23]

PTAL.

This seemed unrelated to the fullscreen patch so I pulled it out to its own CL.

[philipj_slow, 2015-09-29 13:51:58]

lgtm

[dsinclair, 2015-09-29 14:32:46]

This looks like it causes a bunch of test failures. A lot of them are just text
rebaselines where we write out the styles and ::backdrop is in the list now.
There are a few I'm not sure about:


Text says backdrop should not be shown but is.
  dialog/dialogs-with-no-backdrop.html  

Background colour is now grey instead of white.
  dialog/removed-element-is-removed-from-top-layer.html
  dialog/top-layer-containing-block.html
  dialog/top-layer-display-none.html
  dialog/top-layer-nesting.html
  dialog/top-layer-stacking-correct-order-remove-readd.html
  dialog/top-layer-stacking-dynamic.html
  dialog/top-layer-stacking.html


Content moved? I'm wondering if this one isn't related to the change.
  fast/css/pseudo-cache-stale.html


Click missing?
  fast/dom/HTMLDialogElement/modal-dialog-ancestor-is-inert.html


Effecting more elements then expected with style
  fast/css/dynamic-class-pseudo-elements.html


If these are expected changes, let me know. Otherwise I'll start trying to track
down why things are now different.

[dsinclair, 2015-09-30 20:09:11]

On 2015/09/29 14:32:46, dsinclair wrote:
 > Text says backdrop should not be shown but is.
>   dialog/dialogs-with-no-backdrop.html  
> 
> Background colour is now grey instead of white.
>   dialog/removed-element-is-removed-from-top-layer.html
>   dialog/top-layer-containing-block.html
>   dialog/top-layer-display-none.html
>   dialog/top-layer-nesting.html
>   dialog/top-layer-stacking-correct-order-remove-readd.html
>   dialog/top-layer-stacking-dynamic.html
>   dialog/top-layer-stacking.html
> 

For all of the dialog failures it looks like the issue is that the tests do
display:none which no longer overrides the display:block !important.

So, should these just be rebaselined and the display:none on the backdrops
removed? For  dialog/dialogs-with-no-backdrop.html  it looks like just part of
the test is invalid?

[esprehn, 2015-09-30 20:11:32]

What does the spec say? Why !important? I'm not sure it should be forced to
block, maybe it just wants to be forced to an equivalent block display type
in StyleAdjuster?
On Sep 30, 2015 1:09 PM, <dsinclair@chromium.org> wrote:

> On 2015/09/29 14:32:46, dsinclair wrote:
>  > Text says backdrop should not be shown but is.
>
>>    dialog/dialogs-with-no-backdrop.html
>>
>
> Background colour is now grey instead of white.
>>    dialog/removed-element-is-removed-from-top-layer.html
>>    dialog/top-layer-containing-block.html
>>    dialog/top-layer-display-none.html
>>    dialog/top-layer-nesting.html
>>    dialog/top-layer-stacking-correct-order-remove-readd.html
>>    dialog/top-layer-stacking-dynamic.html
>>    dialog/top-layer-stacking.html
>>
>
>
> For all of the dialog failures it looks like the issue is that the tests do
> display:none which no longer overrides the display:block !important.
>
> So, should these just be rebaselined and the display:none on the backdrops
> removed? For  dialog/dialogs-with-no-backdrop.html  it looks like just
> part of
> the test is invalid?
>
> https://codereview.chromium.org/1372413004/
>
> To unsubscribe from this group and stop receiving emails from it, send an
> email to blink-reviews+unsubscribe@chromium.org.
>

To unsubscribe from this group and stop receiving emails from it, send an email
to blink-reviews+unsubscribe@chromium.org.

[esprehn, 2015-09-30 20:11:32]

What does the spec say? Why !important? I'm not sure it should be forced to
block, maybe it just wants to be forced to an equivalent block display type
in StyleAdjuster?
On Sep 30, 2015 1:09 PM, <dsinclair@chromium.org> wrote:

> On 2015/09/29 14:32:46, dsinclair wrote:
>  > Text says backdrop should not be shown but is.
>
>>    dialog/dialogs-with-no-backdrop.html
>>
>
> Background colour is now grey instead of white.
>>    dialog/removed-element-is-removed-from-top-layer.html
>>    dialog/top-layer-containing-block.html
>>    dialog/top-layer-display-none.html
>>    dialog/top-layer-nesting.html
>>    dialog/top-layer-stacking-correct-order-remove-readd.html
>>    dialog/top-layer-stacking-dynamic.html
>>    dialog/top-layer-stacking.html
>>
>
>
> For all of the dialog failures it looks like the issue is that the tests do
> display:none which no longer overrides the display:block !important.
>
> So, should these just be rebaselined and the display:none on the backdrops
> removed? For  dialog/dialogs-with-no-backdrop.html  it looks like just
> part of
> the test is invalid?
>
> https://codereview.chromium.org/1372413004/
>
> To unsubscribe from this group and stop receiving emails from it, send an
> email to blink-reviews+unsubscribe@chromium.org.
>

To unsubscribe from this group and stop receiving emails from it, send an email
to chromium-reviews+unsubscribe@chromium.org.

[dsinclair, 2015-09-30 20:13:07]

On 2015/09/30 20:11:32, esprehn wrote:
> What does the spec say? Why !important? I'm not sure it should be forced to
> block, maybe it just wants to be forced to an equivalent block display type
> in StyleAdjuster?

Spec says block !important.


https://fullscreen.spec.whatwg.org/#user-agent-level-style-sheet-defaults

::backdrop {
  display:block !important;
}

[philipj_slow, 2015-10-01 09:06:19]

On 2015/09/30 20:13:07, dsinclair wrote:
> On 2015/09/30 20:11:32, esprehn wrote:
> > What does the spec say? Why !important? I'm not sure it should be forced to
> > block, maybe it just wants to be forced to an equivalent block display type
> > in StyleAdjuster?
> 
> Spec says block !important.
> 
> 
> https://fullscreen.spec.whatwg.org/#user-agent-level-style-sheet-defaults
> 
> ::backdrop {
>   display:block !important;
> }

Anne (the spec editor) suspects that anything else would lead to some undefined
behavior:
http://krijnhoetmer.nl/irc-logs/whatwg/20151001#l-176

If that doesn't seem to make sense, please file a spec bug at
https://github.com/whatwg/fullscreen/issues/new and mention @upsuper and
@rocallahan to sort this out.

[esprehn, 2015-10-01 09:30:11]

On 2015/10/01 at 09:06:19, philipj wrote:
> On 2015/09/30 20:13:07, dsinclair wrote:
> > On 2015/09/30 20:11:32, esprehn wrote:
> > > What does the spec say? Why !important? I'm not sure it should be forced
to
> > > block, maybe it just wants to be forced to an equivalent block display
type
> > > in StyleAdjuster?
> > 
> > Spec says block !important.
> > 
> > 
> > https://fullscreen.spec.whatwg.org/#user-agent-level-style-sheet-defaults
> > 
> > ::backdrop {
> >   display:block !important;
> > }
> 
> Anne (the spec editor) suspects that anything else would lead to some
undefined behavior:
> http://krijnhoetmer.nl/irc-logs/whatwg/20151001#l-176
> 
> If that doesn't seem to make sense, please file a spec bug at
https://github.com/whatwg/fullscreen/issues/new and mention @upsuper and
@rocallahan to sort this out.

If we don't let you make it display: none then authors are just going to do
something crazy like set top and left or a huge margin to shove the backdrop out
of the way.

::backdrop is not any different than ::before, I don't know what anne is talking
about.

[philipj_slow, 2015-10-01 09:38:08]

On 2015/10/01 09:30:11, esprehn wrote:
> On 2015/10/01 at 09:06:19, philipj wrote:
> > On 2015/09/30 20:13:07, dsinclair wrote:
> > > On 2015/09/30 20:11:32, esprehn wrote:
> > > > What does the spec say? Why !important? I'm not sure it should be forced
> to
> > > > block, maybe it just wants to be forced to an equivalent block display
> type
> > > > in StyleAdjuster?
> > > 
> > > Spec says block !important.
> > > 
> > > 
> > > https://fullscreen.spec.whatwg.org/#user-agent-level-style-sheet-defaults
> > > 
> > > ::backdrop {
> > >   display:block !important;
> > > }
> > 
> > Anne (the spec editor) suspects that anything else would lead to some
> undefined behavior:
> > http://krijnhoetmer.nl/irc-logs/whatwg/20151001#l-176
> > 
> > If that doesn't seem to make sense, please file a spec bug at
> https://github.com/whatwg/fullscreen/issues/new and mention @upsuper and
> @rocallahan to sort this out.
> 
> If we don't let you make it display: none then authors are just going to do
> something crazy like set top and left or a huge margin to shove the backdrop
out
> of the way.
> 
> ::backdrop is not any different than ::before, I don't know what anne is
talking
> about.

If you don't see an implementation problem, then we should just have the spec
changed. I've filed https://github.com/whatwg/fullscreen/issues/25

[philipj_slow, 2015-10-06 08:38:43]

The spec bug was resolved by removing this rule, so nothing to do in this CL :)