At 08.12 -0400 00-09-16, vint cerf wrote:
would it be useful, in the context of establishing peer-to-peer communications
(or even client/server communications) with limited-function mobile devices,
to use SIP as a framework for negotiating the parameters that should guide the
nature of the exchange?
Maybe. The rescap wg is looking into "capabilies" in a different way
(I presume).
I'm thinking, for instance, of a web server that may
usefully discover the functional limits of a mobile before it starts to send
content to that device. The mobile uses SIP to report to the server that it
has X amount of memory, Y amount of display area, color or not, average
data rate it can send or receive, and so on. This information would be used
by the server to configure what it sends to be compatible with the receiving
unit.
perhaps this is an idea that is already being pursued in an IETF
working group?
I think "discovery" of capabilities is an interesting discussion, and
of course it should be dealt with aswell. Some of this information
can be handled by just looking att the tcp control block :-)
Now, I would like, to be honest, this wg to more look at what
stupid(?) design flaws have been made in the existing protocols given
certain capabilities than the actual discovery process. IMAP in
disconnected mode is a good idea, maybe, but might be too chatty.
I.e. long rtt makes chatty protocols like IMAP and SMTP (just as some
examples) quite boring. Can something be done about that (batch
smtp?).
Say we have a very fat pipe between earth and the moon (which current
tcp should be able to handle). Should we run IMAP over that link
which have quite some RTT?
What happens if we have only a 2.4k inmarsat sattelite phone with
900ms rtt, can we still do useful stuff with normal applications
given ppp (the solution is not to change ppp in this discussion)?
paf