W3C home > Mailing lists > Public > www-style@w3.org > September 2011

Re: Suggestion for @global to help with <style scoped>

From: Roland Steiner <rolandsteiner@google.com>
Date: Wed, 28 Sep 2011 15:00:43 +0900
Message-ID: <CACFPSpj1Me57eov9aREcw+o6oXVbksD8GctcTrA1HYf9yZFsKA@mail.gmail.com>
To: fantasai <fantasai.lists@inkedblade.net>
Cc: www-style@w3.org
On Thu, Sep 22, 2011 at 4:30 AM, fantasai <fantasai.lists@inkedblade.net>wrote:

> On 09/21/2011 10:12 AM, Ian Hickson wrote:
>
>> On Tue, 20 Sep 2011, fantasai wrote:
>>
>>> On 09/20/2011 05:22 PM, Ian Hickson wrote:
>>>
>>>> On Tue, 20 Sep 2011, fantasai wrote:
>>>>
>>>>>
>>>>> So, given that, rather than having a "scoped vs. global" switch, how
>>>>> about using a pseudo-class to distinguish whether a portion of the
>>>>> selector is matching out-of-scope elements? E.g.
>>>>>
>>>>>       <style scoped>
>>>>>        section>   h1 { border-bottom: solid; }
>>>>>        :context(body.homepage) h1 { color: red; }
>>>>>        :context(body.archive) h1 { color: gray; }
>>>>>       </style>
>>>>>
>>>>
>>>> ...what element does the :context() match against?
>>>>
>>>
>>> The one defining the scope.
>>>
>>
>> So basically :context(...) is like :scope:matches(... #) ?
>>
>
> Yes. Although I was thinking of having these work differently:
>
>  :context(blockquote) > p { ... }
>  :context(blockquote) p { ... }
>
> So :context(...) would be more like
>
> :scope-or-ancestor-of-scope:**matches(...)
>

That would mean that the same selector that would match an element within
<style scoped> wouldn't match the same element if used in <style> (i.e.,
without scoping), because :scope is then equivalent to :root, and the
context very likely no longer matches. @global doesn't suffer this issue.

But why limit this to the scope? It seems to me, that this could be useful
in a general way, basically A:context(X) saying that the element A must have
an ancestor that matches X, or match X itself. That would also prevent the
above issue, AFAICT. OTOH, it then _is_ the same as "A:matches(X),
A:matches(X #)".
For the purposes of scoped selector matching, we could rule that selectors
within :context (and :matches?) are exempt from scoping (you then couldn't
have a scoped context, but I guess that's a negligible drawback).

As stated elsewhere, I like the parallel between @global <-> querySelector
and no @global <-> queryScopedSelector, but I guess either approach would
work.


Cheers,

- Roland
Received on Wednesday, 28 September 2011 06:01:28 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 17:20:44 GMT