RE: Let's fix THE PROBLEM
"Bob Wyman" <[email protected]> Thu, 17 Jul 2003 19:39:07 -0400
| Newsgroups | gmane.ietf.impp |
|---|---|
| Message-ID | <004b01c34cbc$9df13860$640aa8c0@BOBDEV> |
Joe Hildebrand wrote: > Yes, there are times when I want to publish and subscribe > to a different list of people than the people that are > subscribed to my presence, but that's pub/sub, not presence. Sorry if I'm being dense here, but can you explain why one is "presence" and the other is "PubSub" even though I'm subscribing to "Presence" data in both cases? > I can imagine implementations that use pub/sub for > presence, that don't require anything drastic: What I meant by "drastic" wasn't a technical comment, but rather a process comment. As your example shows, doing Presence over PubSub is pretty obvious and doesn't present major technical problems. The "drastic" thing that I was referring to would be a redefinition of Presence in XMPP to rely on PubSub rather than using the existing mechanism. >I'm not sure how that's a single, monolithic package. Sorry. Bad word choice. A better word might be "stand-alone." In XMPP, Presence doesn't use either PubSub or an events package like in SIP/SIMPLE. This may tend to result in people overloading Presence to accomplish things that would be better done with PubSub since Presence is almost guaranteed to be implemented by everyone whereas PubSub, as an optional protocol extension, is now and probably will remain less commonly implemented. bob wyman [reminder: [email protected] for non-technical discussions, please]