Keith Alexander pointed me to discussions by members of the public-html mailing list on the @profile attribute of HTML's 'head' element. 
I would respond on list, but I am waiting for my W3C HTML WG membership to come through. I have to say I think the W3C process is needlessly complex, especially if they are trying to make the process of joining the HTML WG transparent. While I wait for approval, I will have to resort to posting on my blog instead.
Lachlan Hunt posted on the public-html list:
The profile attribute (which was actually defined for the head element in HTML4) is, in practice, useless.
Not true. The GRDDL specification uses the @profile attribute to specify a method of transforming XHTML in to RDF. GRDDL is supported by RAP (RDF API for PHP) and Jena, the Java-based RDF library.
GRDDL processors can parse data in a more liberal manner. For instance, the GRDDL parser that I wrote (but have not released) does look for the class declarations of various compound microformats (hCard and hCalendar) and uses those to add the relevant parsing stylesheets to the engine if they have not already been declared. But, since domain-specific information may be included on a web page, we cannot rely on GRDDL processors to keep track of all possible semantic formats, nor can we limit the Semantic Web (lower case or upper case, take your pick) to the work done by the microformats community. If a thousand compount microformat-style formats bloom, then a GRDDL processor should be able to do something with them.
@profile allows that to happen. It is non-intrusive - you don't have to use it, but if you do, it makes it easier to scrape data from your site and others in a uniform way.
For more information on GRDDL, see the GetSemantic wiki article.
Microformats defined profile URIs for many of their formats. But in practice, there are many sites that don't bother using them properly, if at all, and more importantly, there are also many tools available that are able to work with the microformats, without even checking if it's present.
Indeed, most sites do not properly use the profile URIs for microformats. If most people didn't brush their teeth, would the British Dental Association stop reccomending it? The fact that tools exist which can parse microformats without profile URIs is no reason to stop using profile URIs. There are tools which require profile URIs to provide data.
So such real world usage and implementation experience indicates that the profile attribute is not necessary, and so it shouldn't be included in HTML5.
No, trotting out the word 'microformats' is no reason to stop using the profile attribute. As for 'real world usage', how can anyone work out what real world usage is? This is the Internet. Netcraft tracks 48m active web servers. Google no longer provides a figure for how large their search index is, but we can guess it's pretty darn huge. How big does 'real world usage' have to be? "Real world usage" is usually a cover for 'I don't use it'.
Karl Dubost wrote:
The profile attribute is, in practice, difficult to use with CMS systems when the *users* want to add more semantics.
That may be so. We can't base tomorrow's HTML on the poorly designed CMS systems of the past. The CMS systems I use allow me to add profile attributes with ease. Hence why my blog has them. Unfortunately, the HTML WG can't specify how people are going to create things.
If we were to take the ability of CMS and authoring tools to produce something, we would have thought CSS 'un-pragmatic' and abolished it a long time ago. Dreamweaver and other authoring tools are still not doing great on the CSS front. But CSS is technically a better way of doing things than tables, and so we persist.
The profile attribute serves a useful purpose. With GRDDL, it allows people to extract domain-specific information from web pages. Cool though microformats are, the road to the Semantic Web has many different routes. Though using class-names like 'vcard' is a good decision for the hCard standard and other microformats, the profile attribute allows other possibilities. Closing off this avenue because it's supposedly not 'practical' would be a tragedy for future development of the Semantic Web.
Tags: html, xhtml, htmlwg, web standards, grddl, semantic web, rdf
