In fact in my bachelor project, the goal (for the moment) is to have a basic client with good serach results on a gnutella network.

When I looked the RFC, it seems better to implement an ultrapeer to work on a search algorithm but I have only 3 months to do the job and in a first phase it seems easier to implement a leaf client than an ultrapeer (I have to learn c# too, because during my studies in telecommunications I coded all the time in C or java).

Due to these constraints, I decided to start first in this direction, developing a leaf with the best search results and depending on the time, I will try to implement the ultrapeer.

So I still have a lot of nights of work ahead me :-))