[standards-jig] FW: Let's fix THE PROBLEM

Bob Wyman bob at wyman.us
Thu Jul 17 23:07:57 UTC 2003


Joe Hildebrand wrote:
> No.  It's more important that we get XMPP finished 
> and out the door as quickly as possible, honoring the 
> spirit of the xmppwg charter, which points us toward 
> maintaining backward-compatibility where possible.
	Yes. I understand. I think it is terribly unfortunate, but I do
understand the motivations and constraints of the current process. The
problem is that it really feels like watching a train wreck in progress.
I'm confident that if PubSub had existed first, then Presence would have
used it. But, now it looks like we're going to see an unnecessary
distinction between the mechanisms for subscribing to and publishing
Presence data and the mechanisms for doing the same for other PubSub
applications... Or, we'll find people trying to squeeze PubSub apps
through the "Presence" protocol since the Presence mechanisms are
guaranteed to be available while PubSub is now and probably will remain
"optional" for the forseeable future... This may be unavoidable based on
current realities, however, I don't believe that it is the "correct"
architectural solution.

> Particular implementations may choose to 
> *implement* presence in terms of pub/sub,
	I really question whether this will actually happen. Anyone who
wants to interwork with existing implementations is going to have to
implement the existing presence mechanism rather than something based on
PubSub. Thus, there won't be a great deal of motivation to move to
PubSub-based Presence. I can see that one might develop a "PubSub view"
of presence, which basically gatewayed between the two ways of accessing
the data, however, this would probably always rely on the basic XMPP
Presence implementation being considered the primary form of Presence
data.
	But, to get the ball rolling here... Would it be reasonable to
submit a JEP that defines "Presence over PubSub"?

		bob wyman




More information about the Standards mailing list