So, from my initial understanding, cjdns nodes are obligated to stay in touch with physically close nodes and address-close nodes.I asked cjd to clarify a bit. "With others there is only one layer, each node makes a decision to route to one of it's directly connected neighbors and that is the end." Whereas cjdns can have routers and switches (non-nodes) in between and still calculate quality links. Calculations are done by number of hops and latency.So, if all computers installed cjdns and became nodes, routers wouldn't need to become nodes themselves. However if routers became nodes, non-node clients of the router would be able to access the cjdns network via NAT66.On Fri, May 24, 2013 at 8:56 PM, Marc Juul <marcjc@gmail.com> wrote:
On Thu, May 23, 2013 at 10:11 AM, Alcides Gutierrez <alcides888@gmail.com> wrote:
Just expanding on my brief email last night:
When using CJDNS you need to find at least one peer to access the network, in this case, Hyperboria. So the router comes preconfigured with peers.
I plan on configuring my laptop to be a node today. If I get it done in time, I'll bring it to the meeting tonight and we can travel to Hyperboria.
Does anyone know how CJDNS actually works? I read the whitepaper and there seems to be no discussion of any of the core concerns of mesh networking protocols. My initial reaction is that it seems like it's a bit of a naive approach, and that the people involved didn't learn from existing projects like BATMAN, OLSR and Babel in designing their protocol. I realize that their goals are different from those projects, but I haven't seen any calculations or simulations of how a large CJDNS network would function. Perhaps CJDNS is just not being developed as a mesh protocol? It seems that the Seattle Meshnet people feel otherwise though, and are actually using it as a meshnet protocol.
Someone please tell me why I/we should care about CJDNS. What problem does it solve that is related to mesh networks? Point to point encryption? Is that it?
--Marc--
Alcides Gutierrez