There is no such requirment. The only difference is in handeling the XML.
Rather than doing this P2P style (with no client taking the role of
server) on of the clients will have to play "server". But ofcourse the
only thing it will "serve" is it's own audiostream, and the only client it
will serve is the other "peer".
Doesnt this really mean than they are both servers each serving the other its audio stream?? By definition that is P2P not client/server.
No, one server will assume the role as "server". The client will ask permission to participate with the conference on the server, sending an IQ request etc. Then after permission SI will be used to create a two way stream for the audio between the server and client. Ofcourse the server will only stream it's own audio input to the client (no need to send the client it's own input back) as long as there are only two people talking. More isn't required, though I can imagine it'd be nice to have when you have the power and bandwith for it.
Peer "To be, or to assume to be, equal". http://dictionary.reference.com/search?q=peer
Well, they're not quite equal as you can see :)
However, supporting this will mean that with the same spec and implementation you can also participate in conferences! Wether those are hosted on a component or on another users machine who has a more advanced client that supports mixing.
Im not saying that we cant design a protocol that will work both ways, that
is the only way we should design it IMO, but there is no need to define in
the protocol that it must be client/server, it should be generic, and
shouldnt restrict doing it still in a p2p fashion for more than two people,
as ive said there is no harm allowing conversations of more than 2 people
using p2p if the bandwidth is available to do it.
This allows 2 people to have a conversation through a direct connection. One is clearly the server, and the other the client. Wether this still matches your definiton of peer to peer I don't know, but it allows to do what you want doesn't it?
The benifit of it is, even if you can't host conferences yourself (for example cause of bandwith etc.) even with the minimum implementation you could still participate in them. Additionally the protocol could define (perhaps just as an extention) a way to manage conferences (for example on a jabber component) remotely. Then, even if I just have a pocket PC, I can still set up a conference, decide wich people to let in / invite / reject, password protect it ect. I can imagine x:data would be used for this (so maybe just field standardization would do it).
_______________________________________________
jdev mailing list
[EMAIL PROTECTED]
http://mailman.jabber.org/listinfo/jdev
