I'm creating a simple gnutella program for a class of mine that simply needs to connect to the gnutella network without having to have any useful functionality, it only needs to connect and receive search requests.

The problem I'm having, however, is that it seems very unclear to me exactly how my program should receive it's initial neighbors. Apparently it used to work by connecting to a well known host that had a list of neighbors to use, but that has changed several times now to UDP Host Caching.

I read the guide at GnuFU en - Gnufu but that gives a generic description of how the method works, whereas i'm looking for a more specific explanation of who to connect to (or send a UDP packet to) in order to receive who my neighbors should be.