Implement the ::backdrop pseudo-element for modal <dialog>.

Implement the ::backdrop pseudo-element for modal <dialog>.

The Fullscreen spec defines ::backdrop, a pseudo-element rendered
immediately beneath top layer elements. This enables the grayed-out
screen effect, for example.

This commit implements ::backdrop for modal dialogs, which are the
currently the only residents of the top layer. The plan is for
fullscreen elements to also use the top layer, so they will use
::backdrop as well.

BUG=253909
Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=153612

[falken, 2013-06-25 14:23:50]

Hi Julien and Elliott

Would you be able to review this?

I'm not totally sure about this approach. This patch makes ::backdrop a child
node of the dialog and it gets reparented just like top layer elements. It may
be possible to keep ::backdrop information entirely in the render tree instead,
but it looked convenient to use the same machinery that ::before and ::after
use. WDYT?

Another issue to keep in mind is event targeting. Now when the ::backdrop is
clicked, the event is dispatched to the dialog. This is how ::before and ::after
also work, but it doesn't seem ideal for ::backdrop. The spec has no comment
about it.

[Julien - ping for review, 2013-06-27 22:49:00]

> I'm not totally sure about this approach. This patch makes ::backdrop a child
> node of the dialog and it gets reparented just like top layer elements. It may
> be possible to keep ::backdrop information entirely in the render tree
instead,
> but it looked convenient to use the same machinery that ::before and ::after
> use. WDYT?

I have some concern about hoisting the PseudoElement around but I don't think
that the approach is bad in itself.

> Another issue to keep in mind is event targeting. Now when the ::backdrop is
> clicked, the event is dispatched to the dialog. This is how ::before and
::after
> also work, but it doesn't seem ideal for ::backdrop. The spec has no comment
> about it.

Talking with Elliott, this is by design of pseudo-elements.

https://codereview.chromium.org/17654008/diff/4001/Source/core/css/resolver/S...
File Source/core/css/resolver/StyleResolver.cpp (right):

https://codereview.chromium.org/17654008/diff/4001/Source/core/css/resolver/S...
Source/core/css/resolver/StyleResolver.cpp:1248: if (pseudoStyleRequest.pseudoId
!= BACKDROP && m_state.parentStyle()) {
It's weird to disallow style cascading for ::backdrop.

https://codereview.chromium.org/17654008/diff/4001/Source/core/dom/Element.h
File Source/core/dom/Element.h (right):

https://codereview.chromium.org/17654008/diff/4001/Source/core/dom/Element.h#...
Source/core/dom/Element.h:576: virtual bool
shouldBeReparentedUnderRenderView(const RenderStyle*) const { return pseudoId()
== BACKDROP || isInTopLayer(); }
Won't this break everything by making backdrop a sibling of the <dialog> instead
of a child? That also means that we make our render tree differ more from the
DOM tree.

https://codereview.chromium.org/17654008/diff/4001/Source/core/rendering/Rend...
File Source/core/rendering/RenderLayer.cpp (right):

https://codereview.chromium.org/17654008/diff/4001/Source/core/rendering/Rend...
Source/core/rendering/RenderLayer.cpp:5629:
posZOrderList->append(toRenderLayerModelObject(backdrop->renderer())->layer());
I haven't seen why it's a fine assumption to assume that ::backdrop will have a
RenderLayer.

https://codereview.chromium.org/17654008/diff/4001/Source/core/rendering/Rend...
Source/core/rendering/RenderLayer.cpp:5630:
posZOrderList->append(toRenderLayerModelObject(child)->layer());
The ordering is wrong as |backdrop| should be *below* the element in the top
layer.

[falken, 2013-06-28 08:06:52]

Thanks for the review!

> Talking with Elliott, this is by design of pseudo-elements.

It makes sense, but it will make it unwieldy to implement
behavior like dismissing the dialog when the outer page
is clicked, or making the dialog "bounce" for attention.
You'd have to detect whether the dialog or its ::backdrop
was clicked.

I'll open a whatwg thread about ::backdrop's
containing block and click events.

https://codereview.chromium.org/17654008/diff/4001/Source/core/css/resolver/S...
File Source/core/css/resolver/StyleResolver.cpp (right):

https://codereview.chromium.org/17654008/diff/4001/Source/core/css/resolver/S...
Source/core/css/resolver/StyleResolver.cpp:1248: if (pseudoStyleRequest.pseudoId
!= BACKDROP && m_state.parentStyle()) {
On 2013/06/27 22:49:00, Julien Chaffraix wrote:
> It's weird to disallow style cascading for ::backdrop.

The spec says ::backdrop "does not inherit from any element
and is not inherited from."
This check should prevent ::backdrop from inheriting from
anything. I think ::backdrop will never be a parent, so
nothing can inherit from it.

It looks like cascading still works fine even with the check.
E.g., the UA stylesheet for ::backdrop background is
overridden in many of my tests.

I've added a comment to the code.

https://codereview.chromium.org/17654008/diff/4001/Source/core/dom/Element.h
File Source/core/dom/Element.h (right):

https://codereview.chromium.org/17654008/diff/4001/Source/core/dom/Element.h#...
Source/core/dom/Element.h:576: virtual bool
shouldBeReparentedUnderRenderView(const RenderStyle*) const { return pseudoId()
== BACKDROP || isInTopLayer(); }
On 2013/06/27 22:49:00, Julien Chaffraix wrote:
> Won't this break everything by making backdrop a sibling of the <dialog>
instead
> of a child? That also means that we make our render tree differ more from the
> DOM tree.

Hm, I was thinking of ::backdrop as a standalone top layer element. So it's
CB would be the ICB like a top layer element. Reparenting it also gets other
probably desired effects like "Ancestor elements with overflow, opacity,
masks, etc. cannot affect it." The spec is not so clear:
"This pseudo-element is a box rendered immediately below the element (and
above the element below the element in the stack, if any), within the same
top layer."

"within the same top layer" sounded to me like ::backdrop is a member of the
stack of elements called the top layer.

Do we want the CB of ::backdrop to be the <dialog> or the ICB? One advantage
of the former is the ::backdrop can be positioned and sized relative to the
dialog. I imagine in most cases the ::backdrop would cover the whole page
though.

https://codereview.chromium.org/17654008/diff/4001/Source/core/rendering/Rend...
File Source/core/rendering/RenderLayer.cpp (right):

https://codereview.chromium.org/17654008/diff/4001/Source/core/rendering/Rend...
Source/core/rendering/RenderLayer.cpp:5629:
posZOrderList->append(toRenderLayerModelObject(backdrop->renderer())->layer());
On 2013/06/27 22:49:00, Julien Chaffraix wrote:
> I haven't seen why it's a fine assumption to assume that ::backdrop will have
a
> RenderLayer.

Done. ::backdrop gets a non-auto z-index now, like top layer elements. I was
relying on forcing ::backdrop to be out of flow positioned to ensure a
layer, but the non-auto z-index is the sure way.

https://codereview.chromium.org/17654008/diff/4001/Source/core/rendering/Rend...
Source/core/rendering/RenderLayer.cpp:5630:
posZOrderList->append(toRenderLayerModelObject(child)->layer());
On 2013/06/27 22:49:00, Julien Chaffraix wrote:
> The ordering is wrong as |backdrop| should be *below* the element in the top
> layer.

It's correct. The layer list is ordered from bottom of the stack to top.

[falken, 2013-06-28 10:54:20]

Anne has clarified the spec. ::backdrop is a first-class citizen in the top
layer, so it's correct to make the renderer a sibling of <dialog>.

[Julien - ping for review, 2013-07-03 01:48:05]

You have addressed my comments and I am happy with the code change (didn't look
at all the tests but the one I did looked fine). Elliott should take an
independent look though.

https://codereview.chromium.org/17654008/diff/28001/Source/core/css/resolver/...
File Source/core/css/resolver/StyleResolver.cpp (right):

https://codereview.chromium.org/17654008/diff/28001/Source/core/css/resolver/...
Source/core/css/resolver/StyleResolver.cpp:1556: || ((e && e->isInTopLayer()) ||
style->styleType() == BACKDROP)
This is now repeated twice, which means a helper function would be nice.

[falken, 2013-07-03 06:14:48]

Julien, thanks for the review. I also added a test for ::backdrop in a
descendant selector.

Elliott, could you review please?

https://codereview.chromium.org/17654008/diff/28001/Source/core/css/resolver/...
File Source/core/css/resolver/StyleResolver.cpp (right):

https://codereview.chromium.org/17654008/diff/28001/Source/core/css/resolver/...
Source/core/css/resolver/StyleResolver.cpp:1556: || ((e && e->isInTopLayer()) ||
style->styleType() == BACKDROP)
On 2013/07/03 01:48:05, Julien Chaffraix wrote:
> This is now repeated twice, which means a helper function would be nice.

Done.

[esprehn, 2013-07-03 23:44:29]

It would be nicer if the ::backdrop was isInTopLayer() too, could we change
isInTopLayer() to be:

bool isInTopLayer() const
{
  if (style()->styleType() == BACKDROP && parentElement()->isInTopLayer())
    return true;
  return isInTopLayer();
}

Then you wouldn't need all the special casing here.

https://codereview.chromium.org/17654008/diff/41001/LayoutTests/dialog/modal-...
File LayoutTests/dialog/modal-dialog-backdrop.html (right):

https://codereview.chromium.org/17654008/diff/41001/LayoutTests/dialog/modal-...
LayoutTests/dialog/modal-dialog-backdrop.html:13: */
Seriously? Crazytown.

https://codereview.chromium.org/17654008/diff/41001/Source/core/css/resolver/...
File Source/core/css/resolver/StyleResolver.cpp (right):

https://codereview.chromium.org/17654008/diff/41001/Source/core/css/resolver/...
Source/core/css/resolver/StyleResolver.cpp:1232: if (pseudoStyleRequest.pseudoId
!= BACKDROP && m_state.parentStyle()) { // The spec disallows inheritance for
::backdrop.
Lets add a method on PseudoStyleRequest which is allowsInheritance() const {
return parentStyle() && pseudoId != BACKDROP; }

btw I don't think this is enough, adjustRenderStyle below is going to muck up
your style based on your parentStyle even though you're being reparented into
the render view. This is a general issue, I think you need to fix
adjustRenderStyle to be smarter for top layer. Lets fix that in a different
patch though, it's only a few cases where you can tell.

https://codereview.chromium.org/17654008/diff/41001/Source/core/css/resolver/...
Source/core/css/resolver/StyleResolver.cpp:1414: return (e && e->isInTopLayer())
|| (style && style->styleType() == BACKDROP);
This looks bad, the PseudoElement for ::backdrop should have isInTopLayer as
true. By design PseudoElements should act like real elements with the same bits
set on them.

https://codereview.chromium.org/17654008/diff/41001/Source/core/dom/Element.h
File Source/core/dom/Element.h (right):

https://codereview.chromium.org/17654008/diff/41001/Source/core/dom/Element.h...
Source/core/dom/Element.h:579: virtual bool
shouldBeReparentedUnderRenderView(const RenderStyle*) const { return pseudoId()
== BACKDROP || isInTopLayer(); }
Why isn't the backdrop in the top layer?

https://codereview.chromium.org/17654008/diff/41001/Source/core/dom/PseudoEle...
File Source/core/dom/PseudoElement.cpp (right):

https://codereview.chromium.org/17654008/diff/41001/Source/core/dom/PseudoEle...
Source/core/dom/PseudoElement.cpp:84: if (!renderer ||
!renderer->style()->regionThread().isEmpty() || renderer->style()->styleType()
== BACKDROP)
Lets remove this and just add an early return below this line:

if (style()->styleType() != BEFORE && style()->styleType() != AFTER)
  return;

[falken, 2013-07-04 08:20:11]

Elliott, thanks for the review.

I changed the patch to make ::backdrop in the top layer on
the DOM side as well.  I went back and forth on the design a
bit.  The advantage is:
+ Less special casing. Don't need changes in NodeRenderingContext or RenderLayer
now.

The disadvantage is:
+ You must add/remove the pseudoelement from the document's
top layer list whenever it's created/destroyed. It's a
little cumbersome and probably slower than letting
RenderLayer insert backdrop in the right place when its
parent is added.

https://codereview.chromium.org/17654008/diff/41001/Source/core/css/resolver/...
File Source/core/css/resolver/StyleResolver.cpp (right):

https://codereview.chromium.org/17654008/diff/41001/Source/core/css/resolver/...
Source/core/css/resolver/StyleResolver.cpp:1232: if (pseudoStyleRequest.pseudoId
!= BACKDROP && m_state.parentStyle()) { // The spec disallows inheritance for
::backdrop.
On 2013/07/03 23:44:29, esprehn wrote:
> Lets add a method on PseudoStyleRequest which is allowsInheritance() const {
> return parentStyle() && pseudoId != BACKDROP; }

Done. That looks a lot better.

> btw I don't think this is enough, adjustRenderStyle below is going to muck up
> your style based on your parentStyle even though you're being reparented into
> the render view. This is a general issue, I think you need to fix
> adjustRenderStyle to be smarter for top layer. Lets fix that in a different
> patch though, it's only a few cases where you can tell.

OK thanks, will take a look.

https://codereview.chromium.org/17654008/diff/41001/Source/core/css/resolver/...
Source/core/css/resolver/StyleResolver.cpp:1414: return (e && e->isInTopLayer())
|| (style && style->styleType() == BACKDROP);
On 2013/07/03 23:44:29, esprehn wrote:
> This looks bad, the PseudoElement for ::backdrop should have isInTopLayer as
> true. By design PseudoElements should act like real elements with the same
bits
> set on them.

Even with the new patch, I have to keep this helper since we enter
adjustRenderStyle as part of createPseudoElementIfNeeded, so the element is
null.

https://codereview.chromium.org/17654008/diff/41001/Source/core/dom/PseudoEle...
File Source/core/dom/PseudoElement.cpp (right):

https://codereview.chromium.org/17654008/diff/41001/Source/core/dom/PseudoEle...
Source/core/dom/PseudoElement.cpp:84: if (!renderer ||
!renderer->style()->regionThread().isEmpty() || renderer->style()->styleType()
== BACKDROP)
On 2013/07/03 23:44:29, esprehn wrote:
> Lets remove this and just add an early return below this line:
> 
> if (style()->styleType() != BEFORE && style()->styleType() != AFTER)
>   return;

Done. Good idea. But we need the !style->regionThread().isEmpty() check still.
It turns out you can have :before or :after and be in a region, in which case
the content shouldn't be generated
(fast/regions/content-flowed-into-pseudo-regions.html)

[esprehn, 2013-07-04 08:34:07]

https://codereview.chromium.org/17654008/diff/61001/Source/core/css/resolver/...
File Source/core/css/resolver/StyleResolver.cpp (right):

https://codereview.chromium.org/17654008/diff/61001/Source/core/css/resolver/...
Source/core/css/resolver/StyleResolver.cpp:1460: return (element &&
element->isInTopLayer()) || (style && style->styleType() == BACKDROP);
If the backdrop is in the top layer, why do you need the BACKDROP check? You
should just remove this method, e && e->isInTopLayer() should work fine since
when you resolve for ::backdrop the parentElement() of the PseudoElement (the
"e" in here) should be isInTopLayer().

https://codereview.chromium.org/17654008/diff/61001/Source/core/dom/Element.cpp
File Source/core/dom/Element.cpp (right):

https://codereview.chromium.org/17654008/diff/61001/Source/core/dom/Element.c...
Source/core/dom/Element.cpp:210: document()->removeFromTopLayer(backdrop);
Lets just do this inside releasePseudoElement(). It seems fine to blindly call
element->document()->removeFromTopLayer(element) there since it's a noop if
they're not in the top layer anyway.

https://codereview.chromium.org/17654008/diff/61001/Source/core/dom/Element.c...
Source/core/dom/Element.cpp:1396: document()->removeFromTopLayer(backdrop);
Yeah lets do this in releasePseudoElement, then you don't need all the
duplication.

https://codereview.chromium.org/17654008/diff/61001/Source/core/dom/Element.c...
Source/core/dom/Element.cpp:2434: document()->removeFromTopLayer(element);
Remove.

https://codereview.chromium.org/17654008/diff/61001/Source/core/dom/ElementRa...
File Source/core/dom/ElementRareData.h (right):

https://codereview.chromium.org/17654008/diff/61001/Source/core/dom/ElementRa...
Source/core/dom/ElementRareData.h:246: m_backdrop = element;
m_backdrop->document()->removeFromTopLayer(...); or do it inside
releasePseudoElement(). :)

[falken, 2013-07-04 09:06:35]

Thanks for the speedy review!

https://codereview.chromium.org/17654008/diff/61001/Source/core/css/resolver/...
File Source/core/css/resolver/StyleResolver.cpp (right):

https://codereview.chromium.org/17654008/diff/61001/Source/core/css/resolver/...
Source/core/css/resolver/StyleResolver.cpp:1460: return (element &&
element->isInTopLayer()) || (style && style->styleType() == BACKDROP);
On 2013/07/04 08:34:07, esprehn wrote:
> If the backdrop is in the top layer, why do you need the BACKDROP check? You
> should just remove this method, e && e->isInTopLayer() should work fine since
> when you resolve for ::backdrop the parentElement() of the PseudoElement (the
> "e" in here) should be isInTopLayer().

createPseudoElementIfNeeded asks for the backdrop's style,
which brings us to StyleResolver::pseudoStyleForElement which
calls adjustRenderStyle with e = 0. So if it's just
e && e->isInTopLayer(), it won't be considered in the top
layer. Once the ::backdrop is created and attached, it keeps
that bad style.

I would think it gets a style recalc when it gets
added to the top layer though, since that forces a reattach.
Will investigate.

Here is a full stack trace:
#0  WebCore::isRenderedInTopLayer (style=0x3cd41e969650, e=0x0) at
../../third_party/WebKit/Source/core/css/resolver/StyleResolver.cpp:1460
#1  0x0000000002d4de2c in WebCore::StyleResolver::adjustRenderStyle
(this=0x3cd41e8886a0, style=0x3cd41e969650, parentStyle=0x3cd41e88a180, e=0x0)
    at ../../third_party/WebKit/Source/core/css/resolver/StyleResolver.cpp:1532
#2  0x0000000002d4cef5 in WebCore::StyleResolver::pseudoStyleForElement
(this=0x3cd41e8886a0, e=0x356f76df0028, pseudoStyleRequest=...,
parentStyle=0x3cd41e851d30)
    at ../../third_party/WebKit/Source/core/css/resolver/StyleResolver.cpp:1305
#3  0x0000000002ac204e in WebCore::RenderObject::getUncachedPseudoStyle
(this=0x3cd41e954038, pseudoStyleRequest=..., parentStyle=0x3cd41e851d30,
ownStyle=0x0)
    at ../../third_party/WebKit/Source/core/rendering/RenderObject.cpp:2806
#4  0x0000000002ac1da0 in WebCore::RenderObject::getCachedPseudoStyle
(this=0x3cd41e954038, pseudo=WebCore::BACKDROP, parentStyle=0x0) at
../../third_party/WebKit/Source/core/rendering/RenderObject.cpp:2776
#5  0x00000000021287a9 in WebCore::Element::createPseudoElementIfNeeded
(this=0x356f76df0028, pseudoId=WebCore::BACKDROP) at
../../third_party/WebKit/Source/core/dom/Element.cpp:2435
#6  0x000000000212450f in WebCore::Element::attach (this=0x356f76df0028,
context=...) at ../../third_party/WebKit/Source/core/dom/Element.cpp:1354
#7  0x0000000001c91f07 in WebCore::Node::reattach (this=0x356f76df0028,
context=...) at ../../third_party/WebKit/Source/core/dom/Node.h:919
#8  0x0000000002124f16 in WebCore::Element::recalcStyle (this=0x356f76df0028,
change=WebCore::Node::NoChange) at
../../third_party/WebKit/Source/core/dom/Element.cpp:1485
#9  0x0000000002125398 in WebCore::Element::recalcStyle (this=0x356f76dd9dd0,
change=WebCore::Node::NoChange) at
../../third_party/WebKit/Source/core/dom/Element.cpp:1552
#10 0x0000000002125398 in WebCore::Element::recalcStyle (this=0x356f76dd9d00,
change=WebCore::Node::NoChange) at
../../third_party/WebKit/Source/core/dom/Element.cpp:1552
#11 0x00000000020ca14c in WebCore::Document::recalcStyle (this=0x356f76dd4028,
change=WebCore::Node::NoChange) at
../../third_party/WebKit/Source/core/dom/Document.cpp:1706
#12 0x00000000020ca49e in WebCore::Document::updateStyleIfNeeded
(this=0x356f76dd4028) at
../../third_party/WebKit/Source/core/dom/Document.cpp:1753
#13 0x00000000020ca66b in WebCore::Document::updateLayout (this=0x356f76dd4028)
at ../../third_party/WebKit/Source/core/dom/Document.cpp:1782
#14 0x00000000020ca850 in
WebCore::Document::updateLayoutIgnorePendingStylesheets (this=0x356f76dd4028) at
../../third_party/WebKit/Source/core/dom/Document.cpp:1826
#15 0x000000000415c9e3 in WebCore::HTMLDialogElement::reposition
(this=0x356f76df0028) at
../../third_party/WebKit/Source/core/html/HTMLDialogElement.cpp:84
#16 0x000000000415cc68 in WebCore::HTMLDialogElement::showModal
(this=0x356f76df0028, ec=@0x7fff02595cbc: 0) at
../../third_party/WebKit/Source/core/html/HTMLDialogElement.cpp:119

[falken, 2013-07-04 12:18:00]

https://codereview.chromium.org/17654008/diff/61001/Source/core/css/resolver/...
File Source/core/css/resolver/StyleResolver.cpp (right):

https://codereview.chromium.org/17654008/diff/61001/Source/core/css/resolver/...
Source/core/css/resolver/StyleResolver.cpp:1460: return (element &&
element->isInTopLayer()) || (style && style->styleType() == BACKDROP);
OK I see now. The problem is just that
StyleResolver::pseudoStyleForElement calls
adjustRenderStyle(state.style(), m_state.parentStyle(), 0);
So adjustRenderStyle for backdrop will always have null "e".

I updated the helper function with a FIXME comment.
I'll try to fix it as part of making adjustRenderStyle
smarter for pseudoelements.

https://codereview.chromium.org/17654008/diff/61001/Source/core/dom/Element.cpp
File Source/core/dom/Element.cpp (right):

https://codereview.chromium.org/17654008/diff/61001/Source/core/dom/Element.c...
Source/core/dom/Element.cpp:210: document()->removeFromTopLayer(backdrop);
On 2013/07/04 08:34:07, esprehn wrote:
> Lets just do this inside releasePseudoElement(). It seems fine to blindly call
> element->document()->removeFromTopLayer(element) there since it's a noop if
> they're not in the top layer anyway.

Done.

https://codereview.chromium.org/17654008/diff/61001/Source/core/dom/Element.c...
Source/core/dom/Element.cpp:1396: document()->removeFromTopLayer(backdrop);
On 2013/07/04 08:34:07, esprehn wrote:
> Yeah lets do this in releasePseudoElement, then you don't need all the
> duplication.

Done.

https://codereview.chromium.org/17654008/diff/61001/Source/core/dom/Element.c...
Source/core/dom/Element.cpp:2434: document()->removeFromTopLayer(element);
On 2013/07/04 08:34:07, esprehn wrote:
> Remove.

Done.

https://codereview.chromium.org/17654008/diff/61001/Source/core/dom/ElementRa...
File Source/core/dom/ElementRareData.h (right):

https://codereview.chromium.org/17654008/diff/61001/Source/core/dom/ElementRa...
Source/core/dom/ElementRareData.h:246: m_backdrop = element;
On 2013/07/04 08:34:07, esprehn wrote:
> m_backdrop->document()->removeFromTopLayer(...); or do it inside
> releasePseudoElement(). :)

Done.

[esprehn, 2013-07-04 16:06:05]

LGTM

[esprehn, 2013-07-04 20:17:49]

https://codereview.chromium.org/17654008/diff/71001/Source/core/dom/Element.cpp
File Source/core/dom/Element.cpp (right):

https://codereview.chromium.org/17654008/diff/71001/Source/core/dom/Element.c...
Source/core/dom/Element.cpp:2450:
document()->addToTopLayer(pseudoElement(BACKDROP), this);
btw, you don't need to call pseudoElement(BACKDROP) here, just use the local
variable element :)

[falken, 2013-07-05 02:16:14]

Thanks for the reviews!

https://codereview.chromium.org/17654008/diff/71001/Source/core/dom/Element.cpp
File Source/core/dom/Element.cpp (right):

https://codereview.chromium.org/17654008/diff/71001/Source/core/dom/Element.c...
Source/core/dom/Element.cpp:2450:
document()->addToTopLayer(pseudoElement(BACKDROP), this);
On 2013/07/04 20:17:50, esprehn wrote:
> btw, you don't need to call pseudoElement(BACKDROP) here, just use the local
> variable element :)

Done.

[commit-bot: I haz the power, 2013-07-05 02:16:40]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/falken@chromium.org/17654008/80003

[commit-bot: I haz the power, 2013-07-05 05:32:02]

Change committed as 153612