Hi all, On 13 April 2010 02:19, Joe Hildebrand <[email protected]> wrote: > > On 4/12/10 3:45 PM, "Stephen Pendleton" <[email protected]> wrote: > >> Microblogging (or blogging in general) is an excellent example though - as >> it is not restricted to IM clients. If you wanted to build a distributed >> non-implementation specific microblogging system that used pubsub, not PEP, >> how would it be done? I wouldn't know how to find your pubsub microblogging >> nodes so I could subscribe to them. However if I always knew your microblog >> was located at node jabberserver.com/nathanfritz (or whatever), or could be >> discovered to be at that node, it would be easy to implement. Without that >> step it is really limited to be used via PEP, which is certainly meant to be >> IM specific, or is limited to a particular implementation (data >> islands=bad). >> >> I think I am just missing a piece of the puzzle here. > > If you want someone to subscribe to your pubsub node, you have to tell them > the JID and node you want them to subscribe to. What that node is is > usually not interesting. For example, see: > > http://xmpp.org/internet-drafts/attic/draft-hildebrand-webdav-notify-00.html > > Which defines a WebDAV property that you can put on a resource, which > specifies the JID and node to subscribe to for changes to that resource. > > If you're just asking how you can have two different microblogs for the same > person with different ACLs, then that's an interesting (and different) > question. We could (for example) define a XEP119-like pointer schema (which > just includes JID+Node, and maybe some descriptive text), and leave those > pointers in the well-defined microblog node. Interested parties could try > to explicitly subscribe to those other JID+nodes if they wanted to.
Good points and good discussion! I am still trying to understand all the use-cases and I think I have started to get at least some of them. Would it be any help if the "http://jabber.org/protocol/disco#items"; -query would include "type" attribute? This way the client could at least show or hide the nodes 'it might be able to understand' without querying all of them. <iq type='get' to='[email protected]/mobile' id='items'> <query xmlns='http://jabber.org/protocol/disco#items'/> </iq> Reply: <iq type='result' from='[email protected]/mobile' to='[email protected]/resource-a' id='items'> <query xmlns='http://jabber.org/protocol/disco#items'> <item jid='somewhere.else.com' node='/geo_loc/[email protected]' name='Geolocation of Pamela Baywatch' type='http://jabber.org/protocol/geoloc' /> </query> </iq> Another point: In overall level I'm not sure if all business cases and problems can be solved in a XEP specification (is it really the place for them?). Please prove me wrong. :-) Cheers, -- tuomas
