Hi Well, I've got this Amiga Gnutella client well on its way now, having done all the network code and most of the routing code. Before I can finish and start...
1346
Tony Kimball
alk@...
Jul 3, 2001 7:35 pm
... Use a PC on the same wire to capture the traffic. I think out-of-sync is an expected result of incompatible protocol implementation, and there is enough...
1347
Serguei Osokine
osokin@...
Jul 3, 2001 8:05 pm
Since I've started writing anyway, I might as well share my recent experience. A few days ago, my postmaster forwarded me a message sent to the abuse@......
1348
Serguei Osokine
osokin@...
Jul 3, 2001 8:13 pm
Hey, does the darn thing swallow all the replies or is it just me? Anyway - Jay, I remember that there was some implementation that took the spec too close to ...
1349
steve bryan
sbryan@...
Jul 3, 2001 9:53 pm
I wish I could remember some details more clearly but I think my overall conclusion was that most out of sync errors I found were probably due to parsing...
1350
Jay Cornwall
jay@...
Jul 3, 2001 11:02 pm
... Hmm, interesting theory. I added some code to show the payload of the previous message whenever the stream became out of sync, and this is what I got from...
1351
Jay Cornwall
jay@...
Jul 3, 2001 11:02 pm
... Well, my connection is a 56k atm, no Ethernet. I don't know of any software on a PC for monitoring serial TCP/IP streams, and which will route it back ...
1352
Jay Cornwall
jay@...
Jul 3, 2001 11:02 pm
... Yep, I was very careful when coding reads across boundaries. Fortunately they don't seem to be occurring just before the stream becomes out of sync (this...
1353
nomead0@...
Jul 5, 2001 3:03 pm
Just a thought... When I first started coding my 'grand experiment' in gnutella the one thing I failed to notice was that the packets must be byte aligned......
1354
steve bryan
sbryan@...
Jul 5, 2001 4:00 pm
... Hmm, I don't know if my observation is similar or conflicting but I found I had to specify packed alignment for the code to work corectly: #pragma options...
1355
Jay Cornwall
jay@...
Jul 5, 2001 5:40 pm
... [..] Ah, I wasn't aware compilers could insert padding into structures as they felt like it. It wasn't the cause of my problems, because although I define ...
1356
ruffnex@...
Jul 6, 2001 3:30 am
Hi Limewire team, Here is a HTTP request field that Limewire often sends: RANGE=bytes=0- My HTTP 206 response includes: Content-Range: bytes 0-3755361/3755362 ...
1357
Christopher Rohrs
crohrs@...
Jul 6, 2001 2:49 pm
... Just to be a stickler, I think you mean "Range: bytes=0-". I only mention this because the Gnutella community has had problems with this header in the ...
1358
ruffnex@...
Jul 6, 2001 4:49 pm
Hi Christopher, ... Apologies, I cut and paste some log output. ... I'm pretty sure I've sent all the data. (see below...) ... LW1.4 seems okay with 206....
1359
Greg Bildson
gbildson@...
Jul 6, 2001 4:54 pm
Simon, Can you show us all of the headers that you send and that you receive? Thanks -greg Ok... Christopher, perhaps you can also shed some light on my ...
1360
ruffnex@...
Jul 6, 2001 5:58 pm
Hi Greg, Hope this helps. Naturally, \r\n are present. Note: -I am ignoring the "Connection: Keep-Alive" field -Send max. 1000000 bytes in message body -I...
1361
Gordon Mohr
gojomo@...
Jul 6, 2001 10:06 pm
Prior metadata discussions have often seemed to assume that any additional metadata in Gnutella queries or query-hits could be squeezed in after a single NUL...
1362
Serguei Osokine
osokin@...
Jul 7, 2001 12:52 am
I've posted a comment to the recent articles about the MusicCity and Gnutella here: http://www.vadem.com/development/gnutella/kazaa.htm Feel free to comment -...
1363
Greg Bildson
gbildson@...
Jul 7, 2001 1:44 am
Serguei, That is a good one. I'm glad you wrote about the flaws in the articles regarding "MusicCity Blows Gnutella Away" and the "Gnutella network can't...
1364
Vinnie
info@...
Jul 7, 2001 1:20 pm
Nice article, Serguei. Actually I'm kinda glad that MusicCity Morpheus is getting that attention. Let THEM sit on the front line!...
1365
Lucas Gonze
lucas@...
Jul 7, 2001 5:35 pm
Hey Oso, It does no good to say that Gnutella has been criticized in idiotic ways, such as the claim that it doesn't scale because inefficiency is quadratic,...
1366
Greg Bildson
gbildson@...
Jul 8, 2001 3:05 am
... From: "Lucas Gonze" <lucas@...> ... Lucas, A lot of what you say is true. However, it is a mistake to say that these improvements are going...
1367
Brown Tiger
browntigerus@...
Jul 9, 2001 11:46 am
Let me get this... If node is firewalled, and sends search query... every other node in search TTL radius reply with results, and forward search result all the...
1368
Vinnie
info@...
Jul 9, 2001 1:04 pm
... I hear what you are saying. But, don't forget that programs like ToadNode will include mediated transfers, thus allowing for file exchanges between...
1369
Vinnie
info@...
Jul 9, 2001 1:34 pm
I have noticed that over 70% of the queries in my query monitor are for complete file names. Which program is sending this out, what is being done to reduce...
1370
Greg Bildson
gbildson@...
Jul 9, 2001 2:45 pm
Even one every 30 seconds would be way too many. Every five minutes might but even that is pushing it. I am suspicious that there may be something other than...
1371
Mike Green
mgreen@...
Jul 9, 2001 4:46 pm
That's why I'm utilizing a connect back (see http://www.cultiv8r.com/cultiv8r/connectback.html). Cultiv8r won't send any PONG if it knows it cannot receive any...
1372
steve bryan
sbryan@...
Jul 9, 2001 6:14 pm
... You are simplifying your statement here, aren't you? Several clients use one hop pings in order to manage their connections so you should respond to these...
1373
Mike Green
mgreen@...
Jul 9, 2001 6:53 pm
Steve, ... If Hop <= TTL = 1, then it still responds with a pong. But not when TTL >= 2 (including the crawlers), which is simply routed. -- Mike...
1374
Greg Bildson
gbildson@...
Jul 9, 2001 10:01 pm
It looks to me that napshare is doing automatic re-querying. If the timeout (600?) can be reset to a low number, then it is a network destroyer. A picture is...
