Re: EC Point Representations

On Feb 12, 2014, at 11:41 PM, Ryan Sleevi <sleevi@google.com> wrote:

> Just to close the loop on this: There's no spec action required at this time.
> 
> If/When the WG introduces additional curves - such as Curve25519 - the spec can be updated to move the ECPoint conversion into handling the key-specific bits.

Following up on this � why wasn�t Curve25519 included in the initial spec? I�m simply curious since a lot of protocols depend on it.

Regards,
NK

> 
> Curve25519 is also something that would prohibit the ECDSA signatures that follow X9.62, so a spec update is required anyways.
> 
> 
> On Fri, Jan 31, 2014 at 11:43 PM, Jim Schaad <ietf@augustcellars.com> wrote:
> As a heads up for those not following the cfrg mailing list at the IETF.
> 
>  
> 
> It is possible that we will need to make the ECPoint typedef be a property of the curve rather than the algorithm name.  There are discussions about the curve Curve25519 and it�s point representation which do not use the X9.62 specification and in fact are specified as being little endian rather than big endian.  (They are looking at only passing the x if I understand things correctly but that is a big if at this point.)
> 
>  
> 
> http://comments.gmane.org/gmane.ietf.tls/11878
> 
>  
> 
> jim
> 
>  
> 
> 

Received on Saturday, 26 April 2014 01:30:48 UTC