On 1 Feb 2004, James Craig Burley wrote:
You might not be aware of this, but the model you want to get *away*
from is the one used by TCP/IP to transmit packets from host to host.
It is true that underlying layer works this way, but most TCP based
services on top of that layer do not.
Put another way, let's say you created a hypothetical email system
(call it dmail, for "direct mail") that provided for no
store-and-forward mechanism at all -- just direct connection from
originating client to receiving client, which is the "simplest" model
along the lines you're proposing
We alrady have this kind of mail system. Its called INSTANT MESSAGING and
we have IETF standards for this system as well as multiple systems of
their own design all operating on the net and quite popular. Taking a
close look at some of the designs for XMPP protocol is good idea when
trying to design replacement for SMTP, I'm not certain we actually want
to design something that parallels that, as in such case it would be
better to simply extend XMPP
Additionally I want to point out that messaging system to replace SMTP
MUST be able to provide support for sending messages between two users who
by design of their software/hardware/network may never be able to talk to
each other directly, that means messages must be able to cross from one
type of network to another (gateway, I know) and this may happen more then
(i.e. A->B->C->D) I can gave you several examples where this situation can
not be avoided (VPN user->office computer->corporate mail server->client
mail server->client computer) and in many other cases we also have networks
(yes full networks) that are not connected to the net all the time and
can't be again by the very design of those nets (direct connections to
mail server for future Mars colony is difficult :)
So to conlude requirement that new NG mail protocol must provide for ability
to transmit messages between non-directly connected locations must be provided
--
William Leibzon
Elan Networks
william(_at_)elan(_dot_)net