Shadow DOM: Toggling class in `.class ::slotted(*)` does not trigger style recalc

• Component: DOM
• Status: RESOLVED FIXED

[Keanu Lee]

Repro: http://jsbin.com/pureji/2/edit?html,output

If I have this style rule:

.class ::slotted(*) {
  background: green;
}

Toggling .class does not cause style recalc.

[Ryosuke Niwa]

*** This bug has been marked as a duplicate of bug 160427 ***

[Antti Koivisto]

This is not a dupe.

[Antti Koivisto]

Created attachment 294518 [details]
patch

[Ryosuke Niwa]

Comment on attachment 294518 [details]
patch

View in context: https://bugs.webkit.org/attachment.cgi?id=294518&action=review

> Source/WebCore/css/StyleInvalidationAnalysis.cpp:114
> +        m_didInvalidateHostChildren = true;

It seems like we don't need to set m_didInvalidateHostChildren if the slot isn't in a shadow tree?
Or is this a kind of optimization to avoid checking whether element is in the shadow tree or not?

> Source/WebCore/css/StyleInvalidationAnalysis.cpp:132
> +        if (shouldCheckForSlots)
> +            return CheckDescendants::Yes;

Don't we need to check the descendants of a 2nd, 3rd, etc... slot elements?
i.e. don't we need to check just !m_ruleSet.slottedPseudoElementRules().isEmpty()?
Put it another way, does this work if a single shadow root had multiple slots
each of which had to be invalidated?

[Antti Koivisto]

(In reply to comment #4)
> Comment on attachment 294518 [details]
> patch
> 
> View in context:
> https://bugs.webkit.org/attachment.cgi?id=294518&action=review
> 
> > Source/WebCore/css/StyleInvalidationAnalysis.cpp:114
> > +        m_didInvalidateHostChildren = true;
> 
> It seems like we don't need to set m_didInvalidateHostChildren if the slot
> isn't in a shadow tree?
> Or is this a kind of optimization to avoid checking whether element is in
> the shadow tree or not?

It is used to avoid O(slots*host children) traversal if there are tons of slots/children.

> Don't we need to check the descendants of a 2nd, 3rd, etc... slot elements?
> i.e. don't we need to check just
> !m_ruleSet.slottedPseudoElementRules().isEmpty()?
> Put it another way, does this work if a single shadow root had multiple slots
> each of which had to be invalidated?

It works because we invalidate all host children when we find the first slot. I don't think optimising more is important at this point.

[Ryosuke Niwa]

Comment on attachment 294518 [details]
patch

View in context: https://bugs.webkit.org/attachment.cgi?id=294518&action=review

>>> Source/WebCore/css/StyleInvalidationAnalysis.cpp:114
>>> +        m_didInvalidateHostChildren = true;
>> 
>> It seems like we don't need to set m_didInvalidateHostChildren if the slot isn't in a shadow tree?
>> Or is this a kind of optimization to avoid checking whether element is in the shadow tree or not?
> 
> It is used to avoid O(slots*host children) traversal if there are tons of slots/children.

Might be worth a comment. Otherwise, someone might come around and *fix* it.

> LayoutTests/fast/shadow-dom/css-scoping-host-and-slotted-context-invalidation.html:81
> +

Can we add a test case with multiple slots in a single shadow tree?

[Antti Koivisto]

Created attachment 294528 [details]
patch for landing

[WebKit Commit Bot]

Comment on attachment 294528 [details]
patch for landing

Clearing flags on attachment: 294528

Committed r208610: <http://trac.webkit.org/changeset/208610>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.

[Radar WebKit Bug Importer]

<rdar://problem/29226652>