[Standards-JIG] JEP-0163: payload type vs. namespace

Ralph Meijer jabber.org at ralphm.ik.nu
Fri Sep 15 09:39:10 UTC 2006


On Wed, 2006-09-13 at 18:36 +0200, Jean-Louis Seguineau wrote:
> Joe could you please illustrate what you mean, pls. Just re-using PSA's
> examples. Thanks

I think he means to use semantic-describing 'namespaces' as node
identifiers. This would be about the same as having a registry of
payload types, where you use URIs as the well-known node identifiers.
People can of then course define their own custom types, too, by using
URIs that are derived from an owned domain name.

Still, I can think of people having more than one blog.

JEP-0163 doesn't mention what to do with the results of the disco items
request sent to a user's JID. The assumption is that the results are all
pubsub nodes. However, a client should check this by sending a disco
info request.

To cover the multiple blog example, I imagine a disco item that
represents a set of pubsub nodes. That disco node would have the
suggested blog namespace as the node identifier. The client would send a
new disco items request to discover the actual pubsub nodes, of which
the names are less important.

Note that this sounds a lot like using pubsub collection nodes with a
well-known name (the semantic describing URI). I am wondering at this
point if we want to define the simplest possible protocol, or the
simplest concept for end-users. I like the ideas of using the roster for
simpler access control in JEP-0163, but I am still not sure about the
presented node discovery process.

Peter and I will meet IRL next week, so this will definitely come up ;-)

-- 
Groetjes,

ralphm




More information about the Standards mailing list