Re: TLS cipersuites for Rserpool

On Monday, September 15, 2003, at 06:44 PM, bill wrote:
> Ok, first by making "the new cypher on the block" the only MUST
> implement we get into a situation where AES might become broken, and
> therefor we have a problem.

This has been discussed fairly extensively in saag, and the
consensus was that AES should be mandatory-to-implement in
new protocols and what to do about 3DES was an open question.

A bigger problem than the possibility of a cipher being
compromised, I think, is that when you've got multiple ciphers
providing different degrees of protection you've got the
possibility of a bid-down attack, which is a protocol attack
and frankly more accessible.  If more than one cipher is
to be specified we need to be explicit about how they're
requested/signaled and to make sure that it's resilient against
that particular type of attack.

Melinda