PIDF for buddylists, was: Re: [Sipping] request to adopt registration package as WG item

I think my comments on using PIDF for the reg package are covered in a
separate email; I'll focus here on the PIDF v. buddylist concept.
However, this is well into simple territory (and IMPP), so I am cc'ing
those lists.

inline:

Paul Kyzivat wrote:
> 
> comments inline.
> 

> > However, note that in cases of notifications of changes, this issue
> > becomes largely moot. Changes in presence of buddies happen one at a
> > time, so that each notify would usually contain just a single presenec
> > doc anyway. THe exception is if the buddylist server is batching the
> > notifies in order to reduce overhead.
> 
> I think batching is potentially an important application of buddylists.
> Buddylist servers may also be a good place to implement filtering to
> further reduce the traffic that
> the subscriber is exposed to. These potentials would be enhanced by
> being able to return the presence of multiple presentities in a single
> presence document.

I agree it would be helpful, although the difference between multipart
and multiple presentities in a single document is not huge. The real
problem is that this is somewhat of a new direction for PIDF late in the
game, and in PIDF was engineered to be minimalistic. Adding support for
a new feature is likely to meet with some resistance. I suppose we could
define an extension to PIDF, although to be honest, from a read of the
schema its not clear to me that a single document with multiple presence
elements is actually disallowed, in whcih case its already there. Of
course, its not likely to be interoperable...


> 
> > > it would then make
> > > sense to subscribe to presence of either presentities or buddylists
> > > without needed to know which is which.
> >
> > There are differences, most notably in the need for diffs in the case
> of
> > buddylist.
> 
> Seems to me these aren't important differences. The features that seem
> important for buddylists and less important for presence of an
> individual are also at least
> potentially useful for the presence of an individual.

Well, the diff feature is critical for buddylist, and currently
undefined for presence. Therefore, we need to have at least a new
package for it. We would also need some new attributes to support the
diffs (a partial/full flag, as we needed for watcherinfo), so we are at
least talking about an extension to PIDF even if multiple presence
elements are allowed. I would be OK with doing that; the difference
between it and multipart is not huge, but its definitely better.

-Jonathan R.


-- 
Jonathan D. Rosenberg, Ph.D.            72 Eagle Rock Avenue
Chief Scientist                         First Floor
dynamicsoft                             East Hanover, NJ 07936
[email protected]                 FAX: (973) 952-5050
http://www.jdrosen.net                  PH:  (973) 952-5000
http://www.dynamicsoft.com


  [reminder: [email protected] for non-technical discussions, please]