Profile Information

Martin Atkins mart at
Wed May 18 06:24:00 PDT 2005

Christopher Schmidt wrote:
> On Wed, May 18, 2005 at 11:13:55AM +0100, Martin Atkins wrote:
 >> ...
> I think that's a good way of doing it: create a minimal description of a
> user, such that it's easy to get data out of it, and return it with the
> OpenID response:
> <Person xmlns=""
>         xmlns:rdf="">
>   <nick>crschmidt</nick>
>   <weblog rdf:resource="" />
> </Person>

I only really latched on to RDF+FOAF because that's what I'm most 
familiar with. I agree with Brad that it could be harmful to depend on 
the flavour of the week and thus exclude people who don't wish to 
embrace such things. If there's something else that's got less emotional 
attachment but retains the extensibility/flexibility of RDF+FOAF I 
wouldn't have any qualms about using it.

The safest solution, as I mentioned in my earlier message, is to not 
mandate anything at all but to give the identity server the ability to 
specify multiple external resources and their types so that the consumer 
can just absorb whichever it prefers. That way I could have both a vCard 
and a RDF+FOAF document and let consumers pick one. We could even go so 
far as to assign a "quality" to each to help choose in the case where a 
consumer supports several.

That's the approach taken by most things on the web: HTML doesn't 
mandate a stylesheet language or image format, it just gives you a 
mechanism to tell the client what you've used and lets the client figure 
out what to do about it. Stylesheets and image formats aren't HTML's 
concern, and likewise profile information isn't OpenID's job.

This does make it harder for consumers, but I presume that pretty 
quickly some de-facto standards will come into play, much like everyone 
uses CSS to style HTML and PNG, GIF and JPEG for images.


(Side-note to any passing microformat advocates: how does one refer to a 
microformat chunk within an HTML document by URI? The easiest way that 
springs to mind is to assign it an ID attribute and use a fragment 
identifier in the URL, but I wonder if you folks have a different way. 
Also, how can one tell what microformat will be encountered without 
fetching and processing the page? MIME types don't seem to apply here.)

More information about the yadis mailing list