W3C home > Mailing lists > Public > www-style@w3.org > April 2010

RE: [css3-mediaqueries][cssom] Empty media queries in stylesheets and the DOM

From: Sylvain Galineau <sylvaing@microsoft.com>
Date: Mon, 26 Apr 2010 14:08:24 +0000
To: Anne van Kesteren <annevk@opera.com>, www-style list <www-style@w3.org>
CC: "Sharon Newman (COHEN)" <sharco@microsoft.com>, Daniel Libby <dlibby@microsoft.com>
Message-ID: <045A765940533D4CA4933A4A7E32597E211F096D@TK5EX14MBXC113.redmond.corp.microsoft.com>
> From: Anne van Kesteren [mailto:annevk@opera.com]

> Right. I'm saying that a CSS3 draft for @media could change it.

Is there one ? Are we planning on one ? In the meantime, we
agree that @media {} is invalid, yes ?

> If you have suggestions for how to clarify it that would help. To me
> it's
> perfectly clear.

:) I'm glad it's clear to you because existing implementations - 3 of them - 
are not exactly helpful on this specific aspect. So either they're all
sloppy or this may not be as clear as it seems. I'd be happy to suggest
changes once I understand what it's supposed to, especially on the OM

> I would like to get full interoperability here, but I think that would
> require moving the MediaList API in the Media Queries specification and
> make it into one model. Or alternatively provide more hooks or
> something
> in the Media Query specification for the API, but I believe neither is
> really desired by the WG as we want to move Media Queries forward.

Which spec the answer goes to is an orthogonal issue to what the interop
runtime behavior is intended to be. I'm really asking about the latter.

This being said, if this is spread across two documents at different
stages then this means that there will be 2+ set of testcases 
for MQ and we will not know whether two browsers interop without running
them both (or all three of them if one includes HTML5's media attribute). 
That's unfortunate. I want MQ to move forward too, but if that means 
punting on OM interop issues then it's worth asking if that's the proper 

More recent modules - Transitions, Transforms and Animations - include the
relevant OM bits. MediaList seems to be a stand-alone interface so it could
fit within this pattern as well. 

> >> (unless a referencing specification says something different, e.g.
> >> <style media=""> is the same as <style media="all"> per HTML5).
> >
> > So:
> > 1. @media {} is invalid
> > 2. If none of the specified queries match then that is equivalent to
> > 'not all'
> > 3. media="" is equivalent to media="all"
> >
> > I think web authors definitely expect #2. But #1 may surprise them in
> > light of #3.
> #3 is also invalid, but needs to be done this way for backwards
> compatibility.

OK, my bad. So media="" is invalid but the absence of the media attribute means
media="all". That one makes sense and seems consistent with no @media
implying @media all {...} and @media {...} being invalid.

> I suppose. I'm open to suggestions and some help in getting it changed
> in all implementations.

Given the correction above, I don't think it's too bad. Setting media="" results
in an invalid query so the content of the style element no longer applies. Removing
the media attribute is equivalent to setting it to "all". There is no real equivalent
in CSSOM i.e. something that removes an @media {...} but I think that's OK. 

Received on Monday, 26 April 2010 14:09:02 UTC

This archive was generated by hypermail 2.4.0 : Friday, 25 March 2022 10:07:45 UTC