hi!without examples of mud_name and player_name, I'd suggest making the domain identify the mud via dns. You do get to route to a component by the full domain ; you could have a jid of [EMAIL PROTECTED] (where player_name and mud-name abide by the character requirements of the respective parts of the jid).
we are thinking about a new protocol for inter-MUD-communication
based on at least the xmpp-core-protocol. a Mud provides communication
facilities to serveral users so we thought about integration of the
jabber server into the mudlib or using a external jabber server.
my favorite would be an external server but then the qeustion is:
do we need a tcp-connection for every user from mud to server or is it possible to use a single connection? in the first case
the jid would be [EMAIL PROTECTED], in the second [EMAIL PROTECTED]/player_name. It it possible to use more than one jid/resource with a single tcp-conn?
Assuming you are hoping to reuse the existing open-source 'jabberd' implementation, I'd mention the option of connecting as a component - its not xmpp-core per-se, but it does let you route all traffic to a particular domain received by jabberd to a particular piece of software over one socket. You run mud-name.server as a part of your mud as a component, and get a inbound flow of remote->s2s->component->mud->player.
It might also be possible to use an external s2s program (I believe there are two now) which acts as a component; a little off the beaten path but requires fewer moving parts. One jabber-server connection can transport data for many users.
-David Waite
_______________________________________________ jdev mailing list [EMAIL PROTECTED] http://mailman.jabber.org/listinfo/jdev
