Shadow DOM: :host() From The First Shadow Context Should Not Style All Shadow Context

• Component: DOM
• Status: RESOLVED FIXED

[Antti Koivisto]

* SUMMARY
The first “:host()” context in a template seems to dictate the the “host” styles for all shadow context in the document. Quoting draft from here https://drafts.csswg.org/css-scoping/#host-selector (if it is up-to-date),:

“The :host pseudo-class, when evaluated in the context of a shadow tree, matches the shadow tree’s host element. In any other context, it matches nothing.”

* STEPS TO REPRODUCE
1. Open included test-case


* RESULTS
Expected: Host 2 from test case shold have a black border instead of pink.
Actual: Both hosts have pink borders.

[Antti Koivisto]

Created attachment 275668 [details]
test

[Antti Koivisto]

<rdar://problem/24668206>

[Antti Koivisto]

Created attachment 275669 [details]
patch

[Andreas Kling]

Comment on attachment 275669 [details]
patch

View in context: https://bugs.webkit.org/attachment.cgi?id=275669&action=review

r=me

> Source/WebCore/style/StyleSharingResolver.cpp:100
> +    if (element.shadowRoot() && !element.shadowRoot()->styleResolver().ruleSets().authorStyle()->hostPseudoClassRules().isEmpty())

This would look so nice if we made authorStyle() return a reference, too.

> Source/WebCore/style/StyleSharingResolver.cpp:287
> +    if (element.shadowRoot() && !element.shadowRoot()->styleResolver().ruleSets().authorStyle()->hostPseudoClassRules().isEmpty())

Mhm.

[Antti Koivisto]

https://trac.webkit.org/r199060

[Alex Christensen]

Comment on attachment 275669 [details]
patch

View in context: https://bugs.webkit.org/attachment.cgi?id=275669&action=review

> Source/WebCore/style/StyleSharingResolver.cpp:288
> +        return nullptr;

nullptr -> false 
:(

[Antti Koivisto]

> nullptr -> false 
> :(

http://trac.webkit.org/changeset/199584