Avoid looping through all attributes for attribute selector.

Avoid looping through all attributes for attribute selector.

An attribute selector may only match a single attribute in a given element
unless the selector matches all namespaces. Break the loop if the selector
has no namespace, or if it has a specific namespace.

R=esprehn@chromium.org
BUG=447832
Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=188250

[esprehn, 2015-01-12 18:13:35]

This isn't going to help html documents because the CSSGrammar.y does:

        $$->setAttribute(QualifiedName(nullAtom, $3, nullAtom),
CSSSelector::CaseSensitive);

when you write [foo] I think?

I think we need to figure out how to make this abort early for html elements so
writing [foo=bar] is faster.

[esprehn, 2015-01-12 18:14:34]

On 2015/01/12 at 18:13:35, esprehn wrote:
> This isn't going to help html documents because the CSSGrammar.y does:
> 
>         $$->setAttribute(QualifiedName(nullAtom, $3, nullAtom),
CSSSelector::CaseSensitive);
> 
> when you write [foo] I think?
> 
> I think we need to figure out how to make this abort early for html elements
so writing [foo=bar] is faster.

Actually hmm, we create it with a nullAtom namespace, you're comparing against
the starAtom. So maybe this does work?

[rune, 2015-01-12 18:34:55]

On 2015/01/12 at 18:14:34, esprehn wrote:
> On 2015/01/12 at 18:13:35, esprehn wrote:
> > This isn't going to help html documents because the CSSGrammar.y does:
> > 
> >         $$->setAttribute(QualifiedName(nullAtom, $3, nullAtom),
CSSSelector::CaseSensitive);
> > 
> > when you write [foo] I think?
> > 
> > I think we need to figure out how to make this abort early for html elements
so writing [foo=bar] is faster.
> 
> Actually hmm, we create it with a nullAtom namespace, you're comparing against
the starAtom. So maybe this does work?

Yes it should, the "slow" path should only hit when you have [*|foo=bar].
Attribute selectors are different from element selector wrt namespaces. [foo] is
the same as [|foo], not [*|foo].

[esprehn, 2015-01-12 18:38:11]

lgtm

[rune, 2015-01-12 21:03:52]

The CQ bit was checked by rune@opera.com

[commit-bot: I haz the power, 2015-01-12 21:04:35]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/844203002/1

[commit-bot: I haz the power, 2015-01-12 22:17:39]

Committed patchset #1 (id:1) as
https://src.chromium.org/viewvc/blink?view=rev&revision=188250