W3C home > Mailing lists > Public > public-html@w3.org > May 2007

Re: The Semantic Debate

From: Benjamin Hawkes-Lewis <bhawkeslewis@googlemail.com>
Date: Mon, 07 May 2007 19:42:26 +0100
Message-ID: <463F7312.2090503@googlemail.com>
To: John Foliot <foliot@wats.ca>
CC: 'Dan Connolly' <connolly@w3.org>, 'Anne van Kesteren' <annevk@opera.com>, www-html@w3.org, public-html@w3.org

John Foliot wrote:

> The current benefit that @role brings to this issue is that it's new
> enough that it has not been (in the words of that working group)
> "polluted" - we can start fresh and clean and get it right.  Using
> @class however has the potential to be ugly - with the emphasis on
> "potential".  Some have cited microformats as an example of reusing
> existing attributes in defense of using @class to assign semantic
> meaning, but what about the "issue" that microformats has with the
> dating scheme they use, deploying the title attribute?
> [http://www.webstandards.org/2007/04/27/haccessibility/]

I'm confused. I thought this was an advantage of the content attribute
rather than the role attribute? How would you put a date inside a role 
attribute?

http://www.w3.org/TR/xhtml2/mod-metaAttributes.html


> @role also allows for scalability via RDF, something not apparent for
> "semantic @class" declarations.  So what are we to do if we have a
> need beyond the prescribed "reserved class declarations"?  The
> proponents of using @class have not answered this question, at least
> I've not seen it in any of the divergent threads that have circled
> around this debate.

This scalability issue worries me, both with HTML5 classes and with 
RDF-defined roles. WHATWG attempts to scale semantics through a process 
of registering new class names (currently on a wiki but the idea was to 
transition to a database with some sort of web service). This simply 
means that you can't count on UAs exposing semantics not predefined in 
the HTML5 spec in a usable manner. I've never worked with RDF, but as 
far as I understand, RDF should in theory allow one to define new roles 
in terms of other roles and interfaces in terms of interfaces. This 
sounds too good to be true. I'm rather suspicious about whether UAs will 
be able to follow a profile URI, read a RDF definition, and construct a 
usable interface. It would sure be nice if we had a working model of a 
UA that can learn new roles, before asserting that we can use it to 
guarantee scalability.

This isn't a problem for the predefined ARIA roles Firefox and AT 
developers are already making use of.

> Well... @role is already "in" Firefox.

Indeed. Which means it's already available cross-platform for free, 
although Firefox's exposure to the Apple Accessibility API needs more 
development.

--
Benjamin Hawkes-Lewis
Received on Monday, 7 May 2007 18:43:28 UTC

This archive was generated by hypermail 2.3.1 : Monday, 29 September 2014 09:38:44 UTC