Support style invalidation for ::content.

Support style invalidation for ::content.

Start supporting style invalidation for selectors containing ::content,
but still use SubtreeStyleChange on distributed children of the content
element.

If you have a selector ".class ::content .target", we will not invalidate
the whole subtree for the element matching ".class", but do style
invalidation for elements matching ".target". We will not cross the
insertion point during invalidation at this point, but merely mark
content elements found during style invalidation with SubtreeStyleChange.

That may be done in a separate patch, but it requires the style
invalidator to have invalidation sets scheduled for elements inside a
shadow tree affect style invalidation of its host subtree.

The addition of :host-context as boundary-crossing and insertion-point-
crossing is in preparation for supporting style invalidation for the
:host-context selector.

This CL is split out of [1].
This CL relies on [2].

[1] https://codereview.chromium.org/639433002/
[2] https://codereview.chromium.org/637273003/

R=esprehn@chromium.org,chrishtr@chromium.org
BUG=424123
Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=183875

[rune, 2014-10-14 12:18:49]

PTAL

[chrishtr, 2014-10-16 02:02:51]

(still processing the rest, sorry for the delay)

https://codereview.chromium.org/652223002/diff/1/Source/core/dom/shadow/Inser...
File Source/core/dom/shadow/InsertionPoint.cpp (right):

https://codereview.chromium.org/652223002/diff/1/Source/core/dom/shadow/Inser...
Source/core/dom/shadow/InsertionPoint.cpp:123: if (change < Inherit &&
styleChangeType() < SubtreeStyleChange)
Is this part all that's needed to fix the attached bug?

[esprehn, 2014-10-16 02:19:06]

You should be able to do this simpler by just making ::content tree boundary
crossing inside the invalidation tree walk. Projected elements must be a subtree
so I don't think you need to use a SubtreeStyleChange, you can just walk the
whole subtree as if it was a /deep/ rule.

https://codereview.chromium.org/652223002/diff/1/Source/core/css/invalidation...
File Source/core/css/invalidation/StyleInvalidator.cpp (right):

https://codereview.chromium.org/652223002/diff/1/Source/core/css/invalidation...
Source/core/css/invalidation/StyleInvalidator.cpp:183: for (size_t i = 0; i <
insertionPoint.size(); ++i)
Just set SubtreeStyleChange on the InsertionPoint itself.

[rune, 2014-10-16 07:23:32]

https://codereview.chromium.org/652223002/diff/1/Source/core/dom/shadow/Inser...
File Source/core/dom/shadow/InsertionPoint.cpp (right):

https://codereview.chromium.org/652223002/diff/1/Source/core/dom/shadow/Inser...
Source/core/dom/shadow/InsertionPoint.cpp:123: if (change < Inherit &&
styleChangeType() < SubtreeStyleChange)
On 2014/10/16 02:02:51, chrishtr wrote:
> Is this part all that's needed to fix the attached bug?

Yes. I started out to support invalidation for ::content, but found this bug on
the way. I can split it into two CLs.

[rune, 2014-10-16 10:58:40]

On 2014/10/16 07:23:32, rune wrote:
>
https://codereview.chromium.org/652223002/diff/1/Source/core/dom/shadow/Inser...
> File Source/core/dom/shadow/InsertionPoint.cpp (right):
> 
>
https://codereview.chromium.org/652223002/diff/1/Source/core/dom/shadow/Inser...
> Source/core/dom/shadow/InsertionPoint.cpp:123: if (change < Inherit &&
> styleChangeType() < SubtreeStyleChange)
> On 2014/10/16 02:02:51, chrishtr wrote:
> > Is this part all that's needed to fix the attached bug?
> 
> Yes. I started out to support invalidation for ::content, but found this bug
on
> the way. I can split it into two CLs.

I split out the bug fix here: https://codereview.chromium.org/637273003/

[rune, 2014-10-16 12:18:24]

On 2014/10/16 02:19:06, esprehn wrote:
> You should be able to do this simpler by just making ::content tree boundary
> crossing inside the invalidation tree walk. Projected elements must be a
subtree
> so I don't think you need to use a SubtreeStyleChange, you can just walk the
> whole subtree as if it was a /deep/ rule.

You mean distribution point crossing? Crossing shadow host/root boundaries is
not necessary for ::content, so making it so will cause unnecessary invalidation
traversal for nested shadow doms.

The problem with following insertion points is that you'd possibly do multiple
passes over the same subtree:

<style>.a .b {}</style>
<div id="a">
  <div>
    <:shadow-root>
      <style>.c ::content .b {}</style>
      <content id="c"></content>
    </:shadow-root>
    <div class="b"></div>
  </div>
</div>
<script>
a.className = "a";
c.className = "c";
</script>


Here, we're scheduling invalidation set for .a on #a and .c on #c. If the
invalidation tree walk moves from #c to .b we'll visit .b two times.

[rune, 2014-10-16 12:18:36]

https://codereview.chromium.org/652223002/diff/1/Source/core/css/invalidation...
File Source/core/css/invalidation/StyleInvalidator.cpp (right):

https://codereview.chromium.org/652223002/diff/1/Source/core/css/invalidation...
Source/core/css/invalidation/StyleInvalidator.cpp:183: for (size_t i = 0; i <
insertionPoint.size(); ++i)
On 2014/10/16 02:19:06, esprehn wrote:
> Just set SubtreeStyleChange on the InsertionPoint itself.

Done.

[chrishtr, 2014-10-16 21:36:58]

lgtm

[rune, 2014-10-17 07:06:28]

The CQ bit was checked by rune@opera.com

[commit-bot: I haz the power, 2014-10-17 07:06:54]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/652223002/40001

[commit-bot: I haz the power, 2014-10-17 07:56:29]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2014-10-17 07:56:30]

Try jobs failed on following builders:
  mac_gpu_retina_triggered_tests on tryserver.chromium.gpu
(http://build.chromium.org/p/tryserver.chromium.gpu/builders/mac_gpu_retina_tr...)

[rune, 2014-10-17 07:59:16]

The CQ bit was checked by rune@opera.com

[commit-bot: I haz the power, 2014-10-17 08:00:19]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/652223002/40001

[commit-bot: I haz the power, 2014-10-17 08:04:25]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2014-10-17 08:04:26]

Try jobs failed on following builders:
  mac_gpu_retina_triggered_tests on tryserver.chromium.gpu
(http://build.chromium.org/p/tryserver.chromium.gpu/builders/mac_gpu_retina_tr...)

[rune, 2014-10-17 08:35:20]

The CQ bit was checked by rune@opera.com

[commit-bot: I haz the power, 2014-10-17 08:35:52]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/652223002/60001

[commit-bot: I haz the power, 2014-10-17 10:29:44]

Committed patchset #4 (id:60001) as 183875