Gnutella Protocoll v0.7 Proposal

is there a link to the GNUXP protocol (v0.7 Peeranha)?

To me, I don't think this 0.7 proposal adds anything, apart from breaking the existing 0.6 protocol and building a new one.

Let me take a few examples to illustrate my point:

1. The 3-way handshake works. It is necessary for Ultrapeer negotiation with gentle redirection of an ultrapeer to leaf status. It is necessary for Gnet traffic compression negotiation.

I understand it can be done with a 2x2-way handshake, but you criticize the 3-way as being complex, so a 4-way is even more complex.

As to simply moving to a 2-way because it is simpler to implement, this is a valid point. However, given the need for 4-way exchanges somtimes, you have to handle exceptions anyway. So let's leave the handshaking as a 3-way process.

2. GUID tagging. This mixes a few concepts. You should have a look at my GGEP "Q" extension proposal, which I have posted on the GDF: It clearly separates between atributes that make sense during a query, and those that make sense during a reply.

Moreover, the "Q" extension is far more extensible that the bits in the GUID.

Finally, don't forget that the GUID is not sent in a query.

3. Renaming of Ultrapeers to something else. Well, I call them Ultranodes. I don't need a protocol 0.7 to call them the way I want. Everyone understand that Ultranodes and Ultrapeers are the same thing.

However, not everything you propose is to throw away. It's just that the premisses of your proposal are wrong, and you target your efforts on things that are superfical inconveniences (but would be a pain to backout) instead of moving forward and constructing.

Live and let learn!

Raphael

Thx master. Here the goals from my old suggestion:
- full HTTP alike connections, to achieve an easy parser
- straight and simple connection sheme, easy and fast
- very flexible desgin for later needs, avoiding a 0.8 in near future *g*
- don't hurt old v0.4 servants, let them still be operable

I won't argue anymore which handshake is more complex and why I don't like the name ultrapeers or the concept behind it. AFAIK the GDF still has no proposal for LAN auto configuration, UDP autofind or internationalisation/UNICODE, is the chat protocol documented meanwhile? v0.7 ideas are about half year old, meanwhile I prefer GNUXP's concept (ask GodXblue for details).
In the past I made suggestions and spend time to improve things and bringing new ideas from different file sharing systems, just take what you like. I found out other developers are not really interested in my experience or support (the friendly way of saying : I know I'm not welcome here). I would have appreciated if Gnutella would be more honest and more cooperative.

Happy developing.

BearShare uses all three handshakes to implement two way challenge/response authentication for host connections.

And a new one that completely handshakes bearshare off of Gnutella!
Good riddance.

Yes but I don´t know if this is the last version...

http://mrgone4662.dns2go.com/forums/...s=&threadid=16

Morgwen

no, it's a very early one

This is the only version I could find, the links from Moak are dead perhaps he has the new version and will offer it again...

Morgwen

Why did you call it "General Gnutella Development Discussion" when there is actually no development??

You complained that it took over 1 year for v0.6 to be spread out on the net? Why do you think the client-developers inmplemented the ability of online-updates in their programs?
Even if a user isn't interested in the new version, he would probably click on "yes, update" just to get rid of the annoying message that a new version is available .
And if you make banners telling about a new protocol version "new, faster protocol v0.7 supported" it would work fine. (faster sounds good for all the average AOL-users , and its a little bit faster anyway )

just my opinion.... nils.lw

The real development of Gnutella is being discussed in the GDF, not here.
This place here is for people who have problems with developing their client and need to ask questions.