On Friday 15 Feb 2013 21:20:33 Mindaugas Gapsevicius wrote: > Dear Matthew, > > first of all thank you for the answers. > May I ask you a yet another question in order to be sure I do not > mix the things up. As far as I understand the darknet uses greedy > algorithm and that wouldn't be a problem with calculated distance > between the nodes. But do you say that the opennet still uses a > random selection of nodes as it was implemented in the initial stages > of freenet? > > Sorry for the lame question, just want to be sure I understand the > evolution of Freenet and current implementation of it.
A node has a single "location", between 0 and 1, randomly assigned on creation. Requests are routed greedily, that is, to the node with the location closest to the key we're trying to find (converted to the same 0...1 space). On darknet, nodes swap locations to make the topology "routable", using an algorithm based on the Metropolis-Hastings algorithm, exploiting the fact that friendship networks are usually small-world networks. On opennet, nodes' locations are constant, but they gain and lose connections when they do requests: When a request completes, the node sending the data may send its noderef back to the requesting node. This passes through many nodes, and can be replaced along the way. If the requesting node decides it wants the connection, it may drop another connection to make space. This is documented on the website, and in the papers section. You might also find the wiki helpful. I'm CC'ing the support list as the explanation is worth other people seeing. > > Best, > miga > > On 31/01/2013 21:07, Matthew Toseland wrote:> On Tuesday 29 Jan 2013 > 11:58:40 Mindaugas Gapsevicius wrote: > >> Thanks Matthew for the answer. Is there any code re visualising darknet > >> available on public domain? It would be very useful to start with. > > > > Not much sorry. A long time ago we used gnuplot or something. Of > course first you'd need to find a darknet - most people use opennet. > <snip> > >>> Interesting idea. There's a lot of data available. We haven't done > much like this since the early attempts to visualise the darknet (a > circle with the locations of darknet nodes and links between them), > based on testnet / swapping. Of course, Freenet is mostly opennet now. > >
signature.asc
Description: This is a digitally signed message part.
_______________________________________________ Support mailing list Support@freenetproject.org http://news.gmane.org/gmane.network.freenet.support Unsubscribe at http://emu.freenetproject.org/cgi-bin/mailman/listinfo/support Or mailto:support-requ...@freenetproject.org?subject=unsubscribe
