Slashdot Mirror
Gnutella v.56 Out?
GnutellaFanBoy writes "If you said Gnutella is dead gnutella.nerdherd.net would like to tell you otherwise. Currently a version .56 is available for download. If your wondering who is still developing gnutella, read this excerpt from their FAQ:
In the interest of protecting those involved, it can only be said that these versions are being produced by someone with access to the source code. No, of course they're not official in the gnullsoft sense, but let's just say that they're legitimate. I would certainly expect more versions, since we've already seen several materialize.
Also of intrest is the protocol specs are out, and people are working on *nix ports. "
Well, it could be, but not always.
When a gnutellaNET search query goes out, it is not transmitted 'peer to peer', ie, directly from you to each of the 2000 other servants. It is routed through gnutellaNET from host to host.
If gnutellaNET is a peer to peer network, so is the internet.
When will this someone make the code available? If the source is not supposed to be out there, floating around, you would think that it benefit from being distributed. Stick it on a few anonymous servers, have a bunch of people archive it on their harddrives, and then eventually, distribute it on via gnutella.
Cthulhu for President!
(darren)
Unfortunately, there are still WAY too many problems with Gnutella.
CSG_SurferDude
LongTail SSH Brute Force analysis tool is here!
It's not very scalable. I don't know what the limitis, I haven't looked at it that hard, but I know that the network will most likely collapse under itself if currently implemented in its current state.
Example:
With 2500 hosts person A connects up to the network, that person sends an INIT broadcast message to the host he's immediately connected to, person B. Person B sends it out to everyone else he's connected to, etc. Then the replies start coming back, single-cast, to person A from 2500 other nodes (or possibly less, due to TTL). Searches happen the same way, except this time with more data.
A couple of problems I see: 1) with more hosts, the number of ping messages sent and the number of ping replies sent will grow (linearly, I think). 2) With an average TTL of 7 it is possible that one node may not be able to reach a segment of the network (any other node that's 8 or more hops away).
The protocol needs to be modified so that there are other ways of getting host information. Maybe a master browser or two. Someone can decide, or be elected a master browser. That client collects a number of hosts, and should a new host connect they can choose to ping the entire network, or talk to the master browser.
Another thing is that it really isn't necessary, as far as I can tell, to ping the entire network to see who's there, other than to possibly get additional hosts to connect to. That could be gotten in much better way. Gnutella has a list of other hosts, why can't it send those to each new person that's connected, rather than them creating a small flurry of activity on the network?
There are other issues that I've run across while developing my gnutella irc bot (gnurd - "/msg gnurd !help" on efnet). I think the developers will work out something, intelligent routing, etc. But in my opinion large changes to the protocol would have to be made to scale up past 10k people or so.
Maybe it's just me, but the GnutellaNet sure seems to be collapsing under it's own weight, and user's greed. I have been running it, but am quickly starting to realize that there isn't much point to it, anymore. It's been over a week since I've had anything successfully download. And over the weekend, at least half a dozen people grabbed a huge mpg off my system, how many copies of it are now available for grabbing - 1. Nice to see that once people leech the stuff, they have no intention of sharing it back out, thus reducing the potential load on my machine.
And last but not least, what's the point anyhow. I know that the pages about the program, claim that it's a tool for filesharing. But be realistic. There are a million easier ways to share files, that are far more reliable. If I have a file to share with a bud, I'll email it to him. Not tell him to log into Gnutella, and grab it. Or I'll toss it onto something like FilesAnywhere. The only reason that I can think of, that people love stuff like this, is it's so much more anonymous than even an FTP site. After all, someone has to be a designated contact for a domain name. On Gnutella, et al, nobody has to identify themselves at all.
"Politicians are interested in people. Not that this is always a virtue. Fleas are interested in dogs." P.J. O'Rourke
...Actually, the protocol specs are at http://capnbry.dyndns.org/gnutella /protocol.html
darren
Cthulhu for President!
(darren)
I wonder how the FSF feels about the GNU name being to promote a piece of software whose authors allege they won't give out its source code until the 1.0 release is done, which now looks exceedingly unlikely ever to happen?
I wonder why Slashdot keeps talking about this software as if it were open source and GNU-happy, when it's not, and may never be?
I wonder if this community is really as strict about its beliefs as it would like to believe, or if it has degraded, as I've long been suspecting, into a what-can-I-get-for-free-as-in-beer crew, just a tiny step above warez d00ds in the ecological chain?
--