Fix usage of pseudo style cache

Fix usage of pseudo style cache

This patch fixes our usage of the pseudo style cache so that we will
use a cached pseudo style when creating pseudo-elements if possible
and so that we cache pseudo styles even if they don't get a renderer.
We also move the logic for having styles with animations or transitions
be non-sharable from RenderStyle::isSharable to the StyleAdjuster, so
that animated pseudo-elements will have their parent's style correctly
marked as non-sharable.

As a result, a pseudo-element without a renderer (e.g. display:none
or content:none) to will correctly return its computed style to
getComputedStyle. This also fixes a regression where the pseudo style
cache can end up very large when its parent style is shared.

BUG=345653
Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=169619

[esprehn, 2014-03-12 05:28:02]

lgtm

[Timothy Loh, 2014-03-12 05:31:03]

The CQ bit was checked by timloh@chromium.org

[commit-bot: I haz the power, 2014-03-12 05:31:07]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/timloh@chromium.org/196143004/20001

[commit-bot: I haz the power, 2014-03-12 07:28:10]

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

[commit-bot: I haz the power, 2014-03-12 07:28:11]

Try jobs failed on following builders: linux_blink_rel

[Timothy Loh, 2014-03-12 09:40:35]

The CQ bit was checked by timloh@chromium.org

[commit-bot: I haz the power, 2014-03-12 09:40:41]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/timloh@chromium.org/196143004/20001

[commit-bot: I haz the power, 2014-03-12 10:16:26]

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

[commit-bot: I haz the power, 2014-03-12 10:16:28]

Try jobs failed on following builders: linux_blink_rel

[Timothy Loh, 2014-03-12 10:18:57]

The CQ bit was checked by timloh@chromium.org

[commit-bot: I haz the power, 2014-03-12 10:19:05]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/timloh@chromium.org/196143004/20001

[commit-bot: I haz the power, 2014-03-12 10:58:06]

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

[commit-bot: I haz the power, 2014-03-12 10:58:07]

Try jobs failed on following builders: linux_blink_rel

[Timothy Loh, 2014-03-12 23:19:14]

The CQ bit was checked by timloh@chromium.org

[commit-bot: I haz the power, 2014-03-12 23:19:24]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/timloh@chromium.org/196143004/20001

[Timothy Loh, 2014-03-12 23:24:58]

The CQ bit was unchecked by timloh@chromium.org

[Timothy Loh, 2014-03-17 04:53:09]

On 2014/03/12 23:24:58, Timothy Loh wrote:
> The CQ bit was unchecked by mailto:timloh@chromium.org

Looks like this is making inspector/elements/edit-dom-actions.html time out. I
think this is caused by the pseudoStyleCache + style sharing interaction
(example below, seems to have always been slow), just exposed for the case where
our pseudo-elements don't get a renderer.

I think for now this doesn't matter too much and I should go ahead and land this
(with the test marked as Timeout) and later we can try and fix the problem
(maybe moving the pseudoStyleCache to hang off Element instead of RenderStyle,
which seems more logical to me anyway). WDYT?

<style type="text/css">
#test { counter-reset: c; }
#test div { counter-increment: c; }
#test div:before { content: counter(c); }
</style>
<div id="test"></div>
<script>
// N^2
for (var i = 0; i < 10000; i++)
    test.appendChild(document.createElement("div"));
</script>

[Timothy Loh, 2014-03-20 00:16:25]

On 2014/03/17 04:53:09, Timothy Loh wrote:
> On 2014/03/12 23:24:58, Timothy Loh wrote:
> > The CQ bit was unchecked by mailto:timloh@chromium.org
> 
> Looks like this is making inspector/elements/edit-dom-actions.html time out. I
> think this is caused by the pseudoStyleCache + style sharing interaction
> (example below, seems to have always been slow), just exposed for the case
where
> our pseudo-elements don't get a renderer.
> 
> I think for now this doesn't matter too much and I should go ahead and land
this
> (with the test marked as Timeout) and later we can try and fix the problem
> (maybe moving the pseudoStyleCache to hang off Element instead of RenderStyle,
> which seems more logical to me anyway). WDYT?
> 
> <style type="text/css">
> #test { counter-reset: c; }
> #test div { counter-increment: c; }
> #test div:before { content: counter(c); }
> </style>
> <div id="test"></div>
> <script>
> // N^2
> for (var i = 0; i < 10000; i++)
>     test.appendChild(document.createElement("div"));
> </script>

esprehn: Could you have a look at my last comment here? Thanks.

[esprehn, 2014-03-20 00:44:48]

Can you explain the n^2 here? Why does this get slow?

I don't think it makes sense to keep the pseudo style cache on Element, we don't
keep the normal style on Element either.

[Timothy Loh, 2014-03-20 04:12:39]

Updated patch as per discussion on IRC.

[esprehn, 2014-03-20 04:17:14]

On 2014/03/20 04:12:39, Timothy Loh wrote:
> Updated patch as per discussion on IRC.

Perfect, lgtm

[Timothy Loh, 2014-03-20 04:18:38]

The CQ bit was checked by timloh@chromium.org

[commit-bot: I haz the power, 2014-03-20 04:18:41]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/timloh@chromium.org/196143004/40001

[commit-bot: I haz the power, 2014-03-20 04:51:02]

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

[commit-bot: I haz the power, 2014-03-20 04:51:02]

Try jobs failed on following builders:
  tryserver.blink on win_blink_rel

[Timothy Loh, 2014-03-20 04:54:17]

The CQ bit was checked by timloh@chromium.org

[commit-bot: I haz the power, 2014-03-20 04:54:21]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/timloh@chromium.org/196143004/40001

[commit-bot: I haz the power, 2014-03-20 05:29:44]

Change committed as 169619