Re: [Cfrg] Point format endian (was: Adoption of draft-ladd-spake2 as a RG document)

[Stephen Farrell <[e-mail]>]

Hi Dan,

On 27/01/15 16:41, Dan Harkins wrote:
>   This spells interop problems even if you just adopt DJB's code to
> handle curve25519. Special cases will be a continued source of trouble.

TBH I think that's a tad overstated.

The main thing that seems to be getting lost here is that big
endian vs. little endian is not a big deal. Either works and
there is little or nothing at stake worth fighting over. I've
expressed my slight preference for LE and it's fine that others
disagree - I'm sure the chairs and/or editors can write text
one way or another and it'll get fixed one way or another
before we're done as folks write code and see that it is better
or worse one way or the other. In the worst possible case,
a very late switch within CFRG from one to the other would
add a line of code and an cranky comment - so long as we get
that done before people deploy stuff widely. (And the stuff
that matters here are specific long term key values and not
code so much. If you think back to the DKIM signature format
changes the same arguments apply and so for ephemeral DH uses
this is something that stays malleable for a while yet.)

Moving on without considering this as a hill to die on and
getting to a decision before deployment but not necessarily
in the next nanosecond seems to me much more important than
which of LE vs BE is chosen.

S.