[Reply-To set to tls(_at_)ietf(_dot_)org.]
On Thu, Jan 01, 2015 at 05:52:55PM +0000, Alexey Melnikov wrote:
On 31 Dec 2014, at 22:14, Nico Williams <nico(_at_)cryptonector(_dot_)com>
wrote:
[Elided here is a sub-thread about how much trouble CFRG has had making
decisions, and how unsuited they are to the task. These were opinions
stated by others. My response was that if CFRG can't choose, that's
fine, let CFRG do what it's good at (cryptology), and let the IETF do
what it's good at (engineering).
For the benefit of ietf(_at_)ietf(_dot_)org readers, the context is that CFRG
was
tasked with producing recommendations for the TLS WG, but CFRG seems
mired in debate about them. From my point of view the risk is that the
logjam won't soon be broken.]
Let the IRTF publish one or more documents describing various curves
suitable for use in Internet protocols. The IETF can pick from among
those.
That is not what TLS WG /SEC AD asked for. They would rather CFRG make
a choice that can be used in TLS and other places, instead of letting
each IETF WG make their own different choice.
We may have to reconsider this then.
If it is really true that CFRG is not adept at making choices, then let
the cryptologists document algorithms (including their cryptographic
attributes, pros, cons, cryptanalysis, general performance analysis,
security considerations, and an overall assessment), and let the
engineers pick from among them. I.e., what we've always done at the
IETF.
This might require some process (a call for consensus in the TLS WG?),
but once done CFRG will be freed to do what it's good at, and to do it
more quickly because there will be no more lengthy arguments about what
to choose. Authors will publish I-Ds, reviewers will review them, and
barring any serious problems, CFRG will progress those I-Ds. The TLS
(and other) WGs can then choose what they like.
I don't mean to start a debate about this _now_. Rather, now is the
time point out that we may have to have this debate, possibly before the
next time we ask the IRTF for recommendations.
Nico
--