[jdev] RE: Atom over XMPP using JEP-0060

[Bob Wyman]

Peter Saint-Andre wrote:
>Hi Bob,
>Has the atom-syntax list gotten closer to resolution on some of these
>issues?
	I don't think so. There are a number of issues that remain to be
resolved before we can come up with a definitive first draft of
"Atom-Over-Jabber"... Of course, it should be obvious that one could use
Jabber File Transfer to move Atom files over Jabber today. However, I think
the more interesting case arises when transferring individual Atom entries
over Jabber as they are created or discovered. (i.e. Like what we're doing
with our experimental system at PubSub.com.)
	The open issues include at least:
	1. In a non-editing environment (when you're not using the Atom API
to add an entry to a blog) is it meaningful or even permitted to have an
Atom:Entry that is *not* wrapped in an Atom:Feed element? The Atom:Feed
element adds a good bit of overhead to the transfer of a single entry. In an
environment like Jabber/XMPP, which provides a means for associating
meta-data with nodes, the Atom:Feed element doesn't really add very much of
value. Ideally, it would be possible to simply pass Atom:Entrys and rely on
Jabber to provide "packaging" for Entries.
	2. How does one associate "feed-level" meta-data with an entry that
is inserted into an aggregate/synthetic feed? Or, if an entry can be passed
without a full atom:feed wrapper (see issue 1), how do you associate some
subset of the feed's metadata with the entry? PaceEntryOrigin is one
proposal being considered for Atom. Another, would be something similar to
the ps:source-feed element that we use for our PubSub Atom over Jabber
experiments. 
	3. If Atom Entries cannot, or should not, be passed without feed
elements, and, if someone has subscribed to a feed of Atom entries
(mini-feeds) using JEP-0060, does it make sense to suffer the overhead of
the "packaging" parts of JEP-0060 messages? Perhaps, rather than having
message\event\items\item\atom it would make sense to simply have
message\event\atom. (i.e. make items/item an optional set of elements that
would only be used when sending more than one item)

		bob wyman