Remove checkForChildrenAdjacentRuleChanges.

Remove checkForChildrenAdjacentRuleChanges.

All uses of SubtreeStyleChange now means strict subtree. All sibling
forest invalidations are done using invalidation sets except on node
insertions and removals.

checkForSiblingStyleChanges now has to invalidate siblings itself on
insertion/removal. Before this change we did a SubtreeStyleChange on a
single element and let checkForChildrenAdjacentRuleChanges mark the
sibling forest for recalc. The reason why we cannot use invalidation sets
when adding/removing nodes, is that we don't change the relevant features
(classes, ids, etc) when we need to figure out. For instance:

<style>
:not(.a) + div { color green }
</style>
<div class="a"></div>
<div>Should be green after insertion</div>

If you insert an element between the two divs, the latter will start
matching the style rule, but we cannot do that with invalidation sets.

Adjustments have been done to the style invalidator to allow scheduling
sibling invalidation sets on SubtreeStyleChange elements, since siblings
will have style recalcs triggered through the invalidation machinery, not
checkForChildrenAdjacentRuleChanges.

BUG=557440
Committed: https://crrev.com/95eaa0cbb6e9a7d348769a335ae702c1214769dc
Cr-Commit-Position: refs/heads/master@{#370097}

[rune, 2015-12-08 09:58:48]

These two need to land before this CL:

https://codereview.chromium.org/1503993002/
https://codereview.chromium.org/1507653002/

[rune, 2015-12-10 00:25:43]

The CQ bit was checked by rune@opera.com to run a CQ dry run

[commit-bot: I haz the power, 2015-12-10 00:27:09]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1509853002/1
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1509853002/1

[rune, 2015-12-10 00:35:23]

rune@opera.com changed reviewers:
+ ericwilligers@chromium.org, esprehn@chromium.org

[rune, 2015-12-10 00:37:58]

https://codereview.chromium.org/1509853002/diff/1/third_party/WebKit/LayoutTe...
File
third_party/WebKit/LayoutTests/fast/css/invalidation/subtree-with-sibling.html
(right):

https://codereview.chromium.org/1509853002/diff/1/third_party/WebKit/LayoutTe...
third_party/WebKit/LayoutTests/fast/css/invalidation/subtree-with-sibling.html:38:
description("A subtree invalidation should not invalidation sibling forest");
should not "invalidate"

[commit-bot: I haz the power, 2015-12-10 02:29:28]

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

[commit-bot: I haz the power, 2015-12-10 02:29:29]

Dry run: Try jobs failed on following builders:
  linux_chromium_rel_ng on tryserver.chromium.linux (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.linux/builders/linux_chromium_...)

[esprehn, 2015-12-10 02:53:52]

Are you sure, for example:

https://code.google.com/p/chromium/codesearch#chromium/src/third_party/WebKit...

    if (computedStyle()->affectedByFocus() &&
computedStyle()->hasPseudoStyle(FIRST_LETTER))
        setNeedsStyleRecalc(SubtreeStyleChange,
StyleChangeReasonForTracing::createWithExtraData(StyleChangeReason::PseudoClass,
StyleChangeExtraData::Focus));
    else if (isElementNode() &&
toElement(this)->childrenOrSiblingsAffectedByFocus())
        toElement(this)->pseudoStateChanged(CSSSelector::PseudoFocus);

that seems to say if we have a FIRST_LETTER style we just do SubtreeStyleChange,
but what if I do :focus + div { } and also have a :first-letter style? Now we'd
only do SubtreeStyleChange and not schedule an invalidation set.

[rune, 2015-12-16 22:52:24]

On 2015/12/10 02:53:52, esprehn wrote:
> Are you sure, for example:
> 
>
https://code.google.com/p/chromium/codesearch#chromium/src/third_party/WebKit...
> 
>     if (computedStyle()->affectedByFocus() &&
> computedStyle()->hasPseudoStyle(FIRST_LETTER))
>         setNeedsStyleRecalc(SubtreeStyleChange,
>
StyleChangeReasonForTracing::createWithExtraData(StyleChangeReason::PseudoClass,
> StyleChangeExtraData::Focus));
>     else if (isElementNode() &&
> toElement(this)->childrenOrSiblingsAffectedByFocus())
>         toElement(this)->pseudoStateChanged(CSSSelector::PseudoFocus);
> 
> that seems to say if we have a FIRST_LETTER style we just do
SubtreeStyleChange,
> but what if I do :focus + div { } and also have a :first-letter style? Now
we'd
> only do SubtreeStyleChange and not schedule an invalidation set.

You're right. Fixed in Patch Set 3.

[rune, 2015-12-17 00:09:24]

https://codereview.chromium.org/1509853002/diff/40001/third_party/WebKit/Sour...
File third_party/WebKit/Source/core/css/invalidation/StyleInvalidator.cpp
(right):

https://codereview.chromium.org/1509853002/diff/40001/third_party/WebKit/Sour...
third_party/WebKit/Source/core/css/invalidation/StyleInvalidator.cpp:216:
pushInvalidationSetsForElement(element, recursionData, siblingData);
Pushing sibling sets here must happen before the early return in this method.

[rune, 2015-12-17 12:55:01]

https://codereview.chromium.org/1509853002/diff/40001/third_party/WebKit/Sour...
File third_party/WebKit/Source/core/css/invalidation/StyleInvalidator.cpp
(right):

https://codereview.chromium.org/1509853002/diff/40001/third_party/WebKit/Sour...
third_party/WebKit/Source/core/css/invalidation/StyleInvalidator.cpp:216:
pushInvalidationSetsForElement(element, recursionData, siblingData);
On 2015/12/17 00:09:24, rune wrote:
> Pushing sibling sets here must happen before the early return in this method.

Done.

[rune, 2015-12-17 12:59:36]

Description was changed from

==========
Remove checkForChildrenAdjacentRuleChanges.

All uses of SubtreeStyleChange now means strict subtree. All sibling
forest invalidations are done using invalidation sets except on node
insertions and removals.

checkForSiblingStyleChanges now has to invalidate siblings itself on
insertion/removal. Before this change we did a SubtreeStyleChange on a
single element and let checkForChildrenAdjacentRuleChanges mark the
sibling forest for recalc. The reason why we cannot use invalidation sets
when adding/removing nodes, is that we don't change the relevant features
(classes, ids, etc) when we need to figure out. For instance:

<style>
:not(.a) + div { color green }
</style>
<div class="a"></div>
<div>Should be green after insertion</div>

If you insert an element between the two divs, the latter will start
matching the style rule, but we cannot do that with invalidation sets.

BUG=557440
==========

to

==========
Remove checkForChildrenAdjacentRuleChanges.

All uses of SubtreeStyleChange now means strict subtree. All sibling
forest invalidations are done using invalidation sets except on node
insertions and removals.

checkForSiblingStyleChanges now has to invalidate siblings itself on
insertion/removal. Before this change we did a SubtreeStyleChange on a
single element and let checkForChildrenAdjacentRuleChanges mark the
sibling forest for recalc. The reason why we cannot use invalidation sets
when adding/removing nodes, is that we don't change the relevant features
(classes, ids, etc) when we need to figure out. For instance:

<style>
:not(.a) + div { color green }
</style>
<div class="a"></div>
<div>Should be green after insertion</div>

If you insert an element between the two divs, the latter will start
matching the style rule, but we cannot do that with invalidation sets.

Adjustments have been done to the style invalidator to allow scheduling
sibling invalidation sets on SubtreeStyleChange elements, since siblings
will have style recalcs triggered through the invalidation machinery, not
checkForChildrenAdjacentRuleChanges.

BUG=557440
==========

[rune, 2015-12-17 13:00:53]

The latest patch set is now based on
https://codereview.chromium.org/1533683002/#ps1

[rune, 2016-01-13 11:58:54]

The CQ bit was checked by rune@opera.com to run a CQ dry run

[commit-bot: I haz the power, 2016-01-13 11:59:08]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1509853002/100001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1509853002/100001

[commit-bot: I haz the power, 2016-01-13 13:33:05]

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

[commit-bot: I haz the power, 2016-01-13 13:33:06]

Dry run: This issue passed the CQ dry run.

[rune, 2016-01-13 16:12:19]

Other patches landed, now ready for review.

[Eric Willigers, 2016-01-19 09:23:39]

lgtm

[rune, 2016-01-19 09:47:04]

The CQ bit was checked by rune@opera.com

[commit-bot: I haz the power, 2016-01-19 09:47:18]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1509853002/100001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1509853002/100001

[rune, 2016-01-19 09:47:22]

On 2016/01/19 09:23:39, Eric Willigers wrote:
> lgtm

thanks!

[commit-bot: I haz the power, 2016-01-19 09:48:43]

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

[commit-bot: I haz the power, 2016-01-19 09:48:44]

Try jobs failed on following builders:
  ios_dbg_simulator_ninja on tryserver.chromium.mac (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.mac/builders/ios_dbg_simulator...)
  mac_chromium_compile_dbg_ng on tryserver.chromium.mac (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.mac/builders/mac_chromium_comp...)
  mac_chromium_gn_rel on tryserver.chromium.mac (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.mac/builders/mac_chromium_gn_r...)

[rune, 2016-01-19 10:14:00]

The CQ bit was checked by rune@opera.com

[rune, 2016-01-19 10:14:00]

The patchset sent to the CQ was uploaded after l-g-t-m from
ericwilligers@chromium.org
Link to the patchset: https://codereview.chromium.org/1509853002/#ps120001
(title: "Rebased")

[commit-bot: I haz the power, 2016-01-19 10:14:18]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1509853002/120001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1509853002/120001

[commit-bot: I haz the power, 2016-01-19 11:52:48]

Description was changed from

==========
Remove checkForChildrenAdjacentRuleChanges.

All uses of SubtreeStyleChange now means strict subtree. All sibling
forest invalidations are done using invalidation sets except on node
insertions and removals.

checkForSiblingStyleChanges now has to invalidate siblings itself on
insertion/removal. Before this change we did a SubtreeStyleChange on a
single element and let checkForChildrenAdjacentRuleChanges mark the
sibling forest for recalc. The reason why we cannot use invalidation sets
when adding/removing nodes, is that we don't change the relevant features
(classes, ids, etc) when we need to figure out. For instance:

<style>
:not(.a) + div { color green }
</style>
<div class="a"></div>
<div>Should be green after insertion</div>

If you insert an element between the two divs, the latter will start
matching the style rule, but we cannot do that with invalidation sets.

Adjustments have been done to the style invalidator to allow scheduling
sibling invalidation sets on SubtreeStyleChange elements, since siblings
will have style recalcs triggered through the invalidation machinery, not
checkForChildrenAdjacentRuleChanges.

BUG=557440
==========

to

==========
Remove checkForChildrenAdjacentRuleChanges.

All uses of SubtreeStyleChange now means strict subtree. All sibling
forest invalidations are done using invalidation sets except on node
insertions and removals.

checkForSiblingStyleChanges now has to invalidate siblings itself on
insertion/removal. Before this change we did a SubtreeStyleChange on a
single element and let checkForChildrenAdjacentRuleChanges mark the
sibling forest for recalc. The reason why we cannot use invalidation sets
when adding/removing nodes, is that we don't change the relevant features
(classes, ids, etc) when we need to figure out. For instance:

<style>
:not(.a) + div { color green }
</style>
<div class="a"></div>
<div>Should be green after insertion</div>

If you insert an element between the two divs, the latter will start
matching the style rule, but we cannot do that with invalidation sets.

Adjustments have been done to the style invalidator to allow scheduling
sibling invalidation sets on SubtreeStyleChange elements, since siblings
will have style recalcs triggered through the invalidation machinery, not
checkForChildrenAdjacentRuleChanges.

BUG=557440
==========

[commit-bot: I haz the power, 2016-01-19 11:52:49]

Committed patchset #7 (id:120001)

[commit-bot: I haz the power, 2016-01-19 11:54:00]

Description was changed from

==========
Remove checkForChildrenAdjacentRuleChanges.

All uses of SubtreeStyleChange now means strict subtree. All sibling
forest invalidations are done using invalidation sets except on node
insertions and removals.

checkForSiblingStyleChanges now has to invalidate siblings itself on
insertion/removal. Before this change we did a SubtreeStyleChange on a
single element and let checkForChildrenAdjacentRuleChanges mark the
sibling forest for recalc. The reason why we cannot use invalidation sets
when adding/removing nodes, is that we don't change the relevant features
(classes, ids, etc) when we need to figure out. For instance:

<style>
:not(.a) + div { color green }
</style>
<div class="a"></div>
<div>Should be green after insertion</div>

If you insert an element between the two divs, the latter will start
matching the style rule, but we cannot do that with invalidation sets.

Adjustments have been done to the style invalidator to allow scheduling
sibling invalidation sets on SubtreeStyleChange elements, since siblings
will have style recalcs triggered through the invalidation machinery, not
checkForChildrenAdjacentRuleChanges.

BUG=557440
==========

to

==========
Remove checkForChildrenAdjacentRuleChanges.

All uses of SubtreeStyleChange now means strict subtree. All sibling
forest invalidations are done using invalidation sets except on node
insertions and removals.

checkForSiblingStyleChanges now has to invalidate siblings itself on
insertion/removal. Before this change we did a SubtreeStyleChange on a
single element and let checkForChildrenAdjacentRuleChanges mark the
sibling forest for recalc. The reason why we cannot use invalidation sets
when adding/removing nodes, is that we don't change the relevant features
(classes, ids, etc) when we need to figure out. For instance:

<style>
:not(.a) + div { color green }
</style>
<div class="a"></div>
<div>Should be green after insertion</div>

If you insert an element between the two divs, the latter will start
matching the style rule, but we cannot do that with invalidation sets.

Adjustments have been done to the style invalidator to allow scheduling
sibling invalidation sets on SubtreeStyleChange elements, since siblings
will have style recalcs triggered through the invalidation machinery, not
checkForChildrenAdjacentRuleChanges.

BUG=557440

Committed: https://crrev.com/95eaa0cbb6e9a7d348769a335ae702c1214769dc
Cr-Commit-Position: refs/heads/master@{#370097}
==========

[commit-bot: I haz the power, 2016-01-19 11:54:01]

Patchset 7 (id:??) landed as
https://crrev.com/95eaa0cbb6e9a7d348769a335ae702c1214769dc
Cr-Commit-Position: refs/heads/master@{#370097}