On 3 Jul 2007, at 04:33, Peter Saint-Andre wrote:
Right. And it seems that perhaps we don't need the version at all. As
far as I can see, the "new" approach (it was suggested by Jacek
back in
2003 so maybe we can't call it new) would have global hashes for
all the
namespaces supported and that's all. Or we could include the service
discovery identity as well.
We still need to do disco, a la current-caps, when we get a first-
time hash, to determine which features map to the hash (and we have
to be very careful to make sure we specify the security
considerations that the receiving client must re-hash the discovered
features to ensure the hashes match).
/K
--
Kevin Smith
Psi XMPP client project leader - http://psi-im.org
