well.....it seems you toooooooooooooo lazy ....anyways thanks...I already did that....I wanted to have a second opinion on it...

Oh sorry, didn't want to be unfriendly. Hey, just post your results here or as a PM and I will take some time and try to fill it with my client experience. I think a compariosn between favourite Gnutella clients, Fasttrack clients and eDonkey could be cool... perhaps wake up some developers. *g*

Greets, Moak

Gnutella will probably never be like Fasttrack or eDonkey. And then again why should it? It's already much better for the one or another task (e.g. for downloading mp3s)

I cant believe that you people are comparing out of date clients with old cars. Clients that are not being developed are draging the network down. Everytime programers want to add a new feature they have to worry about backward compatability. The network will always be dragged down by these dead clients until we get rid of them. Then the network can grow.

For example, all of the modern clients will soon be CORRECTLY implementing swarmed downloads with fule file hashes. The old clients wont have this. Should developers wish to add encryption they cant because these old clients are dead in the water.

Define 'soon'! You mean 'soon' like the GDF stopped arguing whether to use HUGE or GGEP?

But anyways, most of the clients are up-to-date enough, not to cause any serious network troubles. There are some newer clients e.g. Xolox which are not helping the network by sending a lot of queries.

Unregistered: We have standarts, here it's Gnutella protocol v0.4 (+ v0.6 which is just a new handshaking)... also old clients do still following this protocol, great for everyone. If we need a new protocoll, we could introduce it! If you think we need a new protocol which is not backwards compatible, just do it and post detailed suggestions for discussion. While Gnutella clients are in steady improvement, switching over to a new protocoll would be possible.... until now I haven't heard any voice we should leave backwards compatibility. Most ideas like superpeers, hashs, metadata, specialized horizons, swarming and freeloader reducing could be perfecty integrated into Gnutella, it was wisely designed flexible enough. With the new v0.6 handshaking we could do even more crazy things.
Personally I play with the idea of introducing some new descriptors (see older postings about XPING/XPONG etc or packing Query/Queryhits together), not backwards compatible... but it's just design theory yet and far far future.

John: We can compare features anyway. Gnutella allready learned from Fasttrack, superpeer ability will come soon, hashs and metadata will follow. A overview on features, advantages and disadvantages will help coders in design and implementation.

PS: John, define 'a lot of queries', then compare with result from other servents. Plz don't spread false rumours.

Nope, it ain't rumours. Xolox implemented automated requerying. Which is ok, when you do it for a few files, but some users obviously did that for a couple of dozens of files or so. At least I remember that regularly Xolox clients generated the most traffic in my connections tab.

It was also Xolox which used to send you lot's of download requests, once it occupied one upload-slot downloading some divx-movie or so.

I didn't like that...

Ultrapeers are a little different from Fasttrack-Supernodes. The latter actually cache the files clients are sharing and shield their leaves completely from Queries. Ultrapeers do shield their leaves from a portion of the traffic, but not from all of it. You also can connect to more than one Ultrapeer at a time, which you can't do with Fasttrack.

Btw. it's been nearly a year since I read the first articles about Gnutella-Supernodes and whether or not to implement them. Some articles were argueing Gnutella would not remain pure if they implemented any kind of hierarchical structure. That was a couple of months time before I ever heard of fasttrack - I was still using Napster, then. - This is just to illustrate what 'soon' can mean, if you're talking gnutella. And I don't see the GDF trying to speed things up. - But that's just my impression.

oops, that should be 'cache file names' up there...