[tor-dev] RFC on obfs3 pluggable transport

[vmonmoonshine at gmail.com]

>The only issue with your trick, is that I'm not looking forward
>implementing a custom DH key exchange in Python (especially the DH
>parameter generation and public key validation parts).

>From the conversation of Zack with Steven at the breakfast table at
Hotel Cellai, I'm pretty sure that Stegotorus is using DH based on 
elliptic curve and its twist protocol. So you can find an implementation
of it there. Though I haven't looked at the crypto part of the code myself.

Also could you please refer me to the reference that proves that X or
p-X is uniformly random. It seems to me that you are taking a quadratic
residue to an even power, so you always get a bi-quadratic residue for
X. Then if I'm the distinguisher who knows the public group, and I see that the
bi-quadratic residues appears 1/2 times instead of 1/4 of the times then
I can smell that something going on.

Take care,
Vmon