On 2015-11-18 01:27, Mike Jones wrote:
I would strongly argue for using "crv" and "x" to characterize these algorithms.
> Developers will be unpleasantly surprised if we don't and more than likely
unnecessarily confused as well.
I agree. In the case "crv" isn't a curve it would still at least be an
algorithm.
I'm on the fence about whether to use "kty":"EC" or a new key type value,
> such as "EC1" (Elliptic Curve with one coordinate). I'll note that "EC"
> is already designed to allow curves whose representations use
> "x" but not "y", so there's strictly speaking no need for a new key type.
> But I understand the argument that since "x" isn't represented in SEC1
> format for these curves, that a different "kty" value may be appropriate.
IMO, the fact that quite a bunch of popular cryptographic libraries build on
"hardcoded" EC and RSA key-types, motivate a *new name*. It seems that
JWK parsing would be more straightforward as well.
The actual name is fairly unimportant but "EC1" doesn't sound too bad :-)
It fits the algorithms/curves currently on the table, doesn't it?
Anders
-- Mike
-----Original Message-----
From: jose [mailto:[email protected]] On Behalf Of Ilari Liusvaara
Sent: Saturday, November 14, 2015 10:35 AM
To: Anders Rundgren
Cc: [email protected]
Subject: Re: [jose] CFRG ECC in JOSE - thumbprint
On Sat, Nov 14, 2015 at 04:43:27PM +0100, Anders Rundgren wrote:
Hi Ilari,
If these curves are generally recognized as "Edwards" (?) I would
personally prefer that "kty" refer to something Edward-ish like "ED"
although this is (of course) entirely unimportant.
These things are not Edwards curves. These things are abstract public-key
algorithms.
If you truly had Edwards curve over prime field[1], there would be no problem presenting
it in standard "EC" notation, as these things have well-defined curve and both
x and y coordinates that are in Z_p.
And yes, I did consider reusing "crv" and "x", but decided that looks pretty odd (there
is no guarantee that these things are in any way based on elliptic curves, nor that even if they are,
"x" is actually x coordinate of the curve.
[1] But this fails for Edwards curves over prime squared fields, since each
co-ordinate has two subcomponents (AFAIK, higher powers give weak-for-ECC
fields).
-Ilari
_______________________________________________
jose mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/jose
_______________________________________________
jose mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/jose
