Profile Information

[Martin Atkins]

Despite the drawbacks of the approach, using FOAF did have the advantage 
that it simultaneously provided (optionally) profile information about 
the identity being presented. Sites could then use this to populate 
their "name", "email" and "url" fields for the posted comment assuming 
these items were specified.

Where are these to come from now? Sites can hardly display bare URLs in 
the interface as identifiers. Some displayable name needs to come from 
somewhere.

The most obvious approach is to have the user, once they have logged in 
with their Yadis ID, enter a name to use at the site. However, this is 
cumbersome and not very efficient. Also, sites like LiveJournal would 
then presumably create the invisible userid for that ID with the given 
name but the user would have no way to change that name since the hidden 
account won't be able to log in. (there are ways around this, but it 
would be nice if I could keep my profile information in one place and 
have the sites update automatically rather than having to go to each 
site and change my profile.)

Complicating Yadis with profile exchange is almost certainly the wrong 
approach, though.  Now that the complications are hidden away behind a 
pretty URL, perhaps the identity server could optionally return a URL to 
some RDF (which doesn't have to be FOAF, but probably will be for now) 
as well as some kind of identifier pointing at the right item in the RDF 
document.

This still binds Yadis to all that RDF stuff, but it doesn't make any 
preference for the "RDF flavour of the day". Instead, we just say "This 
RDF entity is attached to this identity" and leave sites to do what they 
want with that information, including looking inside the item for FOAF 
attributes if that's what they want to do.

It doesn't really have to be RDF at all; it could easily be just a URL 
to any old document. Once you go down that road, though, you suddenly 
need to deal with one or more of:
     * Content negotiation
     * Specifying multiple documents in the hope that one is acceptable.
     * Consumers having to support a plethora of different and competing
     profile specifications, thus decreasing the likelyhood that Yadis
     will be adopted at all. (won't be able to do anything useful with
     it without *some* kind of profile information)

The nice thing about using RDF is that a new attribute could be added 
which says "this entity uses this URL as its Yadis identifier" so that 
the consumer can find the right RDF entity when lots are present. The 
drawback is that it's a pain to process RDF, and RDF is more complicated 
than most people are willing to deal with.

One of these days I'll manage to write one of these things without 
turning it into an essay...