[microformats-discuss] XFN in Atom

Danny Ayers danny.ayers at gmail.com
Sun Sep 18 10:39:14 PDT 2005

> > As a side note, has there been considerations of using the IANA registry for
> > rel values.
> Not yet, but it's a reasonable proposal to consider.
> To date, W3C has kept the "canonical" list of rel values.  They're adding a
> few in XHTML2, e.g. rel="profile", which we're thinking of making explicit
> even independent of XHTML2 (i.e. before it progresses all the way to REC).

Point of info - although Atom has in effect reused the (X)HTML <link>
construct, it's defined slightly differently in Atom Format than in
the HTML specs. In brief, the "rel" attribute may either be a IRI
(i.e. internationalized URI) or a take one of the (simple string)
values registered with IANA - initially: "alternate", "related",
"self", "enclosure", and "via".

So (if the XFN namespace owners were comfortable with this) a
possibility might be :



(that's a HTML version of draft-ietf-atompub-format-11 - still waiting
for the RFC)

I don't recall this being discussed, but as Atom contains a Person
construct already, it might make sense to reuse that outside of its
usual contexts (author or contributor). There would also be the
possibility of using XFN to create a "Structured Extension", see:


Yet another approach might be to include the XFN/XHTML data inside the
content part of an Atom entry. A possible complication is that I don't
think there's any explicit spec-defined "preferred" place for the
profile URI, generally XHTML content will go in a <div> so <head>
isn't available.

Presumably it wouldn't be unreasonable to use something like:
<atom:link rel="http://www.w3.org/1999/xhtml#profile"
href="http://gmpg.org/xfn/11" />

Alternately I think it would be ok to include a whole XHTML document
as the content, giving the type attribute as "application/xhtml+xml"
rather than just "xhtml". That way the profile URI could be included
as an attribute of <head> following the usual microformat approach.




More information about the microformats-discuss mailing list