[tor-dev] Fwd: Re: [tor-talk] Bridge: Why not just stateless TCP socket proxy / forwarders?

[Roger Dingledine]

On Mon, Jan 16, 2012 at 12:53:38PM +0100, Fabio Pietrosanti (naif) wrote:
> So, if a third party would like independently to:
> - develop a TCP forwarder (very simple  code)
> - submit to the BridgeDB
> - decide to which host to connect back
> 
> which would be the step do be done from technical standpoint of view?
> 
> Because that way it would be possible for a lot of third party to
> develop very lightweight "Tor TCP Proxy" that doesn't have inside other
> than the basic logic to do the TCP Proxy.
> 
> Is the availability for third party software/script for that goals
> considered?

I know some research groups who are working on various parts of this
problem. But one piece that I believe nobody is working on right now
is a mechanism for having bridgedb hear about other addresses that it
should give out as bridges.

Right now bridgedb only learns about bridges via the networkstatus and
bridge descriptor files it imports from the bridge directory authority;
and it also assumes that whenever it learns about a bridge it also
learns the identity fingerprint for the bridge (so it can decide what
distribution strategy bucket to put it in).

If somebody wants to grab this problem and brainstorm a todo list,
maybe with the help of Aaron and Karsten (the last sighted bridgedb
maintainers :), that'd be grand.

--Roger