W3C home > Mailing lists > Public > w3c-wai-er-ig@w3.org > April 2001

Re: WAI class use, attribute quoting, and name capitalization

From: Sean B. Palmer <sean@mysterylights.com>
Date: Mon, 2 Apr 2001 16:48:29 +0100
Message-ID: <015001c0bb8e$788f9600$14da93c3@z5n9x1>
To: <w3c-wai-er-ig@w3.org>, "Harvey Bingham" <hbingham@acm.org>, "Leonard R. Kasday" <kasday@acm.org>
Cc: <wai-xtech@w3.org>
> What would you think of using RDF to indicate
> 1. that the class is informationally relevent for the user  (i.e.
> more than just appearance)
> 2. A description of what the class means.

Yes, this is a good use of RDF, and it is something that we have been
modelling in the past (using XPath). I suppose that classes already
have URIs, but it is quite difficult to map between them when using
XPath, unless you say:-

   mydoc:[@class='myclass'] = yourdoc:[@class='yourclass'] .

But where would this assertion be stored? This is the problem with
XHTML - it has a terrible lack of linking or extension mechanisms.
Hence, I suggested that if people just use a URI for the class data,
you could quite easily infer that:-

   mydoc:[@class='URI1'] = yourdoc:[@class='URI1'] .

Where the URIs are the same, because that's a fundamental axiom of
URIs: "sameness". Thus, you wouldn't need to make this declaration,
saving us from the annotation problem. Note that Internet Explorer 6.0
Beta supports character escaping in CSS, and so it is now possible to
style these URI classes, as well as give them meaning.

My other suggestion was using Modularization. It occured to me that
instead of using class="chapter" in the WAI specs, why not make a
DTD/XSD for it, and then mark it up with <wai:chapter>? Using XHTML
classes where we could (should?) be using bona fide markup seems like
a bit of a hack to me, although it might not be worth going to the
trouble of creating an entire XHTML module (which is quite a big
project) for the WAI. Then again, what else is m12n going to be used
for?

> I had once suggested adding at "alt" attribute to style sheets to
> convey the meaning of a class, but it ran afoul of the philosophy
> that style sheets should contain no content.

Actually that's no big problem because CSS should be an XML language
(it was developed before XML existed...) and so there is no easy way
to point inside it. I would have no problem with adding semantics to
CSS, but I think its easier working with the styling hooks in the
markup.

--
Kindest Regards,
Sean B. Palmer
@prefix : <http://webns.net/roughterms/> .
:Sean :hasHomepage <http://purl.org/net/sbp/> .
Received on Monday, 2 April 2001 12:04:25 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:01:32 UTC