Re: Re: Using the DHT to find download sources

"M. Peterson" <[email protected]>
Newsgroups gmane.network.gnutella.devel
Message-ID <[email protected]>
Hi

I wounder what you discuss here. It is obvious, that a DHT is a better
search than the gntuella protocol.
There has been already posted a link to a study and as well deverlopers from
essential gnutella projects say this.
So you want a DHT additionally or instead? You want it for all files or rare
files or popular files?
kademlia is for c++ a very good one and mijito maybe for java (next to
yacy.net).

we we should change the list here to [email protected]
Mojito promised a very good Finding of buddies with the DHT, but is has
never been implemented and rises the question, if rare files (e.g. a buddy)
are really found in the DHT).

If you are looking for a Instant Messenger using a DHT to find friend, see
here: http://retromessenger.sf.net
We are currently searching for cpp/h and wixwidget coders.

The store of a Hash in the DHT is another aspect, do you think Bitzi could
be done in a DHT?
There are always limitations, e.g. http://offsystem.sf.net is the new
followup of torrent, emule and gntuella, and it has link http-links to
identify the files.
Please test it and tell us, if such a long url could be stored in a DHT?

Spam you never can avoid in a peer-to-peer network, but you can in a
friends-to-friends network.
As this is exclusive connecting to trusted friends only.

This allows as well some forwarding of information, e.g. bitzi-links, in the
messenger there is already a function to rate and forward links.
http://retromessenger.sourceforge.net/images/RetroMessengerLinkCloud2.png
that is free of spam, or you get it over a trusted node, which has
auto-forwarding of links.
But even then there is a rating of the links.

Furthermore see how off-links are organized, it is the next generation of
filesharing, you do not share the physical data, but only the metadata the
off-links.
Gnutella is such vunerable for downloading physical chunks, esp. if they are
copyrighted ones.

My question is therefore more, can in a DHT any off-url stored. yes of
course, we have ww.yacy.net. which is doing this for normal urls or could do
this as well for urls not browsble with mozilla (but with offsystem).

So we have all we need, and why then should gnutella be used? it is
oldfashioned and not worth to update, as it is there. even a DHT for
metadata is there.

The only interesting question is, how can you share metadata urls like
offurls or bitzi urls or ed2k urls not in a p2p network, but a f2f network.
So they are swapped from trusted friend to trusted friend, libretroshare,
the library behind retromessener is doing this. Currently each friends need
to click to forward ne urls to all friends.
That could be automated, but then there is more spam or you have to look at
the ratingof urls.

So here a DHT is only used to boostrap a serverless friend to friend
network, over which in secure f2f connections metadata urls are passed
around.

Every node gets that file or url, as it is tiny. So we have a paradigm
change:
from p2p to f2f
from physical chunks to metadata-urls
from gntuella to DHT
from DHT peers to DHT only for boostrapping usage.
and, last not least, a change of the paradigm from search to push.

One friend is pushing the url to the blocks from one friend to another
friend.
That should define gnutella new, and not a DHT.

To organize pushed, we need a f2f network based, and that is libretroshare,
we currently develop with wxwidgets gui framework. As it is based (soon) o
PGP keys to find the friend in the DHT, this is basic.
and a Java gui might be settled on it to, so we have Qt and Wxwidgets, MFC
is missing and java or web.

So the conclusion must be: a Push into a trusted friend is better than a DHT
search on any node.
Isn´t it? And that both should define the_gdf2.

Regards Max


[Non-text portions of this message have been removed]
lmpx.com only provides a reader for public news (NNTP) servers. It is not affiliated with the servers or forums shown here and is not responsible for the content of articles, which is written by their respective authors.