... Definitely an HTTP. ... I'm glad we have you onboard. We need the full range of opinions so that everyone can listen at all the arguments and ponder them....
Happy new year! In the past couple of months I have learned a tremendous amount by reading the posts in this group. Thanks, all! I read the QRP proposal dated...
4147
Raphael Manfredi
Raphael_Manfredi@...
Jan 1, 2002 7:30 pm
I understand modern clients use special GUIDs idenfified as: GUID[8] = 0xff; GUID[15] = 0; /* 0 for 0.4, or 1 for 0.6 */ And that this special GUID is used as...
4148
Raphael Manfredi
Raphael_Manfredi@...
Jan 1, 2002 7:31 pm
From BearShare 2.4.0b7, I get the following handshaking headers in the "standard" part (i.e. not in the second part, after I have acknowledged the handshaked...
LimeWire, BearShare, and other "modern" clients use this technique for all message GUIDs (pings and queries). We do not use it for the server (client) GUID,...
4150
Raphael_Manfredi@...
Jan 1, 2002 11:00 pm
Thanks for the fast reply, Christopher. ... Could you clarify the following points: 1. Does GUID[15] need to be "1" to mean "understands ping/pong reduction"? ...
4151
Raphael Manfredi
Raphael_Manfredi@...
Jan 2, 2002 12:20 am
This afternoon, I had some free time and was away of any computers, so I spent some time reading Gnutella litterature. In particular, I read both Mike Green's...
I actually dropped the ping/pong reducing/caching in its entire from my client, reason being that I have seen issues with clients that do have them. -- Mike ...
... Like Raphael Manfredi, I disagree that we should model Gnutella after HTTP etc... We SHOULD use fixed-length fields, why? Because it conserves BANDWIDTH,...
Thank you for your comments on my proposal! And sorry for taking a week to reply, as I've been busy with the holidays and all... ... Yep, and it helps conserve...
... Why do we need to move so quickly? Shouldn't we try to get this protocol done RIGHT rather than too hastily? ... I think one other issue that is of great...
... there ... download ... as often ... has a ... a client ... more often ... file ... to the ... requesting ... the ... Hmm, I hadn't thought of using a...
4157
Justin Chapweske
justin@...
Jan 2, 2002 2:08 pm
... My problem with using vanilla 2169 is that there will be a large amount of misses on GETs like the above. The nice thing about CAW is that all of these...
... started ... simply, ... GNUtella ... always ... Thanks, I knew I probably wasn't the first to think or something like this. As for adverse affects, I'm...
... Perhaps we should compare notes. Thank god we never released it to the public (its only in the beta). I was extremely skeptical of an .exe that only...
... This is implemented exactly as per the LimeWire ping/pong scheme. ... BearShare has a yet unused feature where it can be told the largest hops value it is...
... get ... the ... large ... We want it as right as we can get it. But time is essential now. Currenlty BS Private Data, LW Meta Info Query, LW Meta Info...
... The server side must at least give a unique GUID for the queued file. The client should save this GUID as if it were the Servent ID in a sent PUSH request....
... do you think needs to be added? Hmm, that was a rather old message Justin. I had made a mistake with mixing the CAW and RFC docs up. CAW allready provides...
Vinnie, ... I don't want to tell you what or what you should not include with your applications, but I've learned from the mistake of including 3rd party ...
Hmmm. Interesting idea. Although you don't really want to reduce the experience of the average user - there are plenty of alternative software packages....
This is a brief summary of the Hops-Flow message implemented in BearShare 2.4.0: function code: 0x2f (=47 decimal) payload: version: 1 byte, unsigned integer ...
4167
Raphael_Manfredi@...
Jan 2, 2002 11:51 pm
... Whatever the merits of that idea are, I find if formidable that you: * Never discussed something touching the protocol before implementing it. * Introduced...
... It sounds like a rather dangerous feature to me. But I am really wondering about the Hops for a Hops-Flow message. Is this Hops set by the connecting node?...
Well I guess I will stop being a voy on this discussion forum and lay something in. It seems we have a lot of contridiction and here and there ideas lately. ...
... Flow, ... wondering ... connecting ... forwarded ... through the ... Hops-Flow messages are dropped if ttl>1 or hops>0. They are not forwarded or...
... <snippage> ... The only problem, as I see it given the behavior described, is what should be in the initial header of the 0.6 handshake. As I understand...
... Well, take a look at <http://groups.yahoo.com/group/the_gdf/files/Development/Gnutella%200.6%20Handshaking%20Protocol>, especially the example below "3....
... .. [snip] .. ... I must admit that this idea is neat. Just wish you'd thrown this into the GDF before implementing it, but allas, no obligations. I like...
4174
Raphael Manfredi
Raphael_Manfredi@...
Jan 3, 2002 9:11 pm
The LimeWire ping/pong reduction paper talks about estimating the horizon in the Appendix. It says that hosts exchange their H tables regularily. How is this...
