[104685] in Cypherpunks
Re: Java applet security, exportability, Jon Postel haiku
daemon@ATHENA.MIT.EDU (Anonymous)
Mon Oct 26 21:32:28 1998
Date: Tue, 27 Oct 1998 02:41:31 +0100
From: Anonymous <nobody@replay.com>
To: drc@adni.net, coderpunks@toad.com
Reply-To: Anonymous <nobody@replay.com>
> 1a) Am I correct that this is proof against the man in the middle?
Not as long as your MITM can force a sabotaged applet on your client
(contains altered g^x, uses bad RNG code, sends a copy in the clear for
attacker's convenience, etc).
...
> 2) Is 160 bits sufficient for y?
No -- although I don't know much about the difficulty of the DLP with
various parameters, I do know the KEA spec uses 160 bits to (allegedly)
force 2^80 operations upon attackers, not enough to keep your kid sister
out.
> 3) Is K = SHA1(k) a suitable way to generate the session key?
If SHA-1 is a good hash, probably so.
> 4) What symmetric cypher would you recommend I use?
Arc4 has legal isues, among other things, and IDEA is patented, leaving
3DES as the only one remaining of your choices. (Or, if you're looking for
a moderately old cipher, there's always... :)
