On 05/10/2008 11:37 AM, Sylvain Hellegouarch wrote:
Therefore in a web page you might want to have both:
<link rel="related" href="xmpp://foo"
label="Chat about foo" />
<link rel="discuss" href="xmpp://foo"
label="Chat about foo" />
I go farther. For example, in a post about administering the Ejabberd
XMPP server you might have the following links:
<link rel='discuss'
href='xmpp:ejabberd@xxxxxxxxxxxxxxxxxxxx?join'
label='Realtime chatroom about ejabberd'/>
<link rel='discuss'
href='http://www.ejabberd.im/forum'
label='Web forum about ejabberd'/>
<link rel='related'
href='http://en.wikipedia.org/wiki/Ejabberd'
label='Wikipedia page about ejabberd'/>
There are two http: URIs in this post, but with radically different
purposes -- doing some background reading vs. interacting with real live
human beings who might be able to answer your questions. You can't
differentiate between the two http: URIs based on the scheme, but you
can based on the rel. I must admit that I don't understand why there is
such resistance (at least by one person on this list) to differentiating
between those purposes via the rel attribute. But I'm going to write an
I-D about this and follow the procedures defined in Section 7.1 of RFC
4287, which in turn reference RFC 2434. Those procedures do not involve
review by the Atompub WG (which in any case has been terminated) nor by
the subscribers to this residual mailing list. We'll see if the IESG
agrees with my argument that the "discuss" relation is a reasonable
addition to the registry of link relations. If so, I'll see you all
again during IETF Last Call!
Thanks for the feedback.
Peter