[standards-jig] On the Subject of Avatars

Evan Prodromou evan at prodromou.san-francisco.ca.us
Mon Jul 21 17:32:36 UTC 2003

>>>>> "MM" == Matt Mankins <mankins at media.mit.edu> writes:

    MM> Given that pubsub does have a way to store data, and that its
    MM> use will likely be more and more coupled with the server, it
    MM> seems that pubsub actually IS the simplest because:

    MM> * -does not- require an extra component on the server side,
    MM> or deployment of a new server 
    MM> * -reuses- pubsub
    MM> * -includes- all other user information (vCard) (err, assuming
    MM> that someone someday creates pubsub nodes for user information)

So, this would make sense if it was in the future tense -- will not,
will reuse, will include -- but not in the present tense. You're
assuming that in the future that PubSub is made final, that it's
implemented in servers, and that a future standard (as of yet
unproposed) puts vCard into PubSub.

The Dead Simple Avatars proposal is _entirely_upward_compatible_ with
a future implementation depending on PubSub. If we use DSA, and later
vCards are made pubsub items, then avatars naturally follow along.

Right now, we have in common use three count 'em three user
info protocols:

            * jabber:iq:register (parts go into the vcard with jabberd 1.4.x)
            * vCard
            * jabber:iq:search AKA JUD

JEP 0084 will introduce a FOURTH protocol for user information -- not
designed around any of the other three, and without an upgrade path
for the other three to work towards.

I agree that pubsub has a bright shining future. BUT, I think we need
to recognize the real problems with user information in Jabber, and
move towards a unified design for user information. JEP 0084 is a step
in the wrong direction.


Evan Prodromou
evan at prodromou.san-francisco.ca.us

More information about the Standards mailing list