> I assume that either the server or the client would have to send some kind
> of message to initiate the key exchange negotiation?  So there would be
> a backwards compatibility issue here, in the event that the other side
> does not understand the key exchange request message.  In that case it
> would be good if the node could fall back on an unencrypted connection.

I would assume that encrypted connections would be specified using a
different address in something like the new multilayered addressing scheme
I brought up. So we would have dh:tcp/localhost:19114 for encrypted
connections using DH Nothing before tcp means use the default, which
would be unencrypted. I'm not at all sure if this is the best way to do
things, but it seems okay. Another way would be to have a handshake
exchange in which the node could say what encryption methods it
understood. This is undesirable in situations when you don't want people
to know that you're running a Freenet node, so you eschew contact with
other Freenet nodes. You want to 1) make connections in straight encrypted
mode and 2) have any nodes that find out about you from DataSource make
connections to you in straight encrypted mode.

> socket layer data handling versus the higher layer.  SSL for example
> handles this by specifying a different port number for the "secure"
> versions of the protocols.  This makes sure there is no conflict where

This is a really great solution. Too bad it won't work for us. We could
have even port number secure and odd port numbers insecure. Just
kidding. That's silly. :-)



_______________________________________________
Freenet-dev mailing list
Freenet-dev at lists.sourceforge.net
http://lists.sourceforge.net/mailman/listinfo/freenet-dev

Reply via email to