[foaf-dev] FOAF AutoDiscovery vs RDF AutoDiscovery

Dan Brickley danbri at danbri.org
Mon May 5 13:13:52 BST 2008

Julian Bond wrote:
> Peter Ansell <ansell.peter at gmail.com> Mon, 28 Apr 2008 16:03:58
>> 2008/4/26 Julian Bond <julian_bond at voidstar.com>:
>>>  There seems to be a growing movement towards using XRDS for a more generic
>>> auto-discovery. See XRDS-Simple[1] for instance.
>> Has it been used with RDF before? Any prominent examples?
> It's all new, so not yet.
>>>  Has anyone thought about getting FOAF in there as well so that FOAF become
>>> just another service type within a single XRDS file?
>> Why not create the same protocol in RDF so you don't have to have a
>> centralised list of types that can be used?
> Feel free to re-invent it!
> Seriously though, we're looking at a Use Case where a user is signing up
> to a new social network using OpenID. As part of that process we'd like
> to pick up their FOAF file to populate their profile form and find
> members already on the service that appear in their foaf:knows
> relations. Since there is almost certainly an XRDS-Simple document for
> OpenID discovery in that process, it makes some sense to also pick up
> the FOAF file location at the same time from the same document. What
> we're proposing is an entry something like this.
> <Service priority="1">
>   <Type>http://xmlns.com/foaf/0.1/</Type>
>   <MediaType>application/rdf+xml</MediaType>
>   <URI>http://www.voidstar.com/foaf.xml</URI>
> </Service>
> Would the FOAF community have any objection to using the FOAF xmlns in
> this circumstance given the commitment at the top of
> http://xmlns.com/foaf/spec/ ?

(Which commitment exactly?)

But sure, it seems worth defining a way to play in this universe (even 
if we secretely harbour the suspicion that it might all be easier and 
cooler with SPARQL etc).

I'd prefer you used  <Type>http://xmlns.com/foaf/spec/</Type> rather 
than keep '0.1' floating around in URLs unnecessarily. The spec has been 
updated in place all these years, so whatever version we're on now, 0.1 
is not really it. Now if the <Type> URI will be used as anything other 
than an opaque naming string or pointer to more info, then maybe the 0.1 
version is better, but my understanding of XRDS suggests this will be OK 
using /spec/ ...

Thanks for chasing this up,




More information about the foaf-dev mailing list