[Cryptech Tech] Requirements for signing the Tor consensus

[Daniel Cegiełka]

2014-03-05 18:22 GMT+01:00 Linus Nordberg <linus at nordberg.se>:
> Joachim Strömbergson <joachim at secworks.se> wrote
> Wed, 05 Mar 2014 13:07:01 +0100:
>
> | > Does this match the use case: http://ed25519.cr.yp.to/python/sign.py>
> | > (So I know the operations to be supported.)
> |
> | I threw together a repo shell for developing the core. Do you have a
> | pointer to ed25519 in Tor to see exactly how it is called (data types etc)?
>
> 1. Tor directory authorities don't use Ed25519 yet. No such proposal
> exists.
>
> 2. Tor relays don't use Ed25519 yet. Proposal 220 [0] describes how they
> might do that in the future. I know there's a rewrite of that proposal
> sitting somewhere not published yet, so don't make decisions based on
> it.
>
> 3. Tor relays do use Curve25519 for the "ntor" handshake since
> 0.2.4.8-alpha (2013-01-14). See [tor-spec] section 5.1.4. for
> details. This application requires high performance.

Yes, I saw the Curve25519 in Tor repo but I don't remember Ed25519 in
src. I am currently using Ed25519 (signify) to sign packages in my
small linux distro.

btw. referring to Curve25519:
http://safecurves.cr.yp.to/

Daniel