Le 10/18/12 5:47 AM, Dhruv Chopra a écrit :
You may want to look in to "SIP". At the very least, when a client is up
and running - it needs to send a special REGISTER message containing its
name/id to the server. On receiving this the server would associate that
name/id with the source IP address. After this the client needs to send
periodic keep-alives or else the server can delete the mapping.
I recently worked on UDP and you also would like to read about "UDP
punching" if your clients are going to be behind NATs.
In any case, as soon as the client send an UDP message to the server,
the problem will be solved.
The server could also broadcast messages, all the client should receive
them, and dscard them if they are not interested in the message, but
that means the server should have a session, which it currently does not
have...
--
Regards,
Cordialement,
Emmanuel Lécharny
www.iktek.com
