Hey Ilari!

Thanks for the speedy review : )

Inline:

On Wed, Aug 30, 2023 at 2:10 AM Ilari Liusvaara <[email protected]>
wrote:

> On Wed, Aug 30, 2023 at 01:27:49AM +0000, Michael Jones wrote:
> > Orie Steele<https://twitter.com/OR13b> and I have written a new
> > specification creating algorithm identifiers for JOSE and COSE that
> > fully specify the cryptographic operations to be performed -
> > something we'd promised to do during our presentation to the JOSE
> > working group<
> https://datatracker.ietf.org/meeting/117/materials/slides-117-jose-fully-specified-algorithms-for-jose-and-cose-00
> >
> > at IETF 117. The introduction to the specification (quoted below)
> > describes why this matters.
> >
> > The specification is available at:
> > *
> https://www.ietf.org/archive/id/draft-jones-jose-fully-specified-algorithms-00.html
>
> Two small things that came up in quick read:
>
> - Why ES25519/ES448 instead of Ed25519/Ed448? Or is having an alg and
>   crv with the same name an issue for some implementations?
>
>
IIRC there are examples of using the same label for multiple registry
entries, especially in COSE:

kty: WalnutDSA, alg: WalnutDSA
kty: HSS-LMS, alg: HSS-LMS

- https://www.iana.org/assignments/cose/cose.xhtml

Personally, I don't think this is a good thing to do.

AFAIK, there are no JOSE cases where this happens.

  I would associate ES25519/ES448 with certain other stuff...
>
>
> - Ed25519 and Ed448 are fully specified algorithms, so the description
>   field could just be "Ed25519"/"Ed448".
>
>
I agree that they are fully specified.
I still think it's better to not name the curve and the algorithm with the
same value, but perhaps the pattern from COSE registries is "good" / worth
continuing ?


>
>
> Then:
>
>
> "Discuss the treatment of EDCH-ES and its ephemeral keys."
>
> There is actually an related issue in JOSE and COSE.
>
> ECDH-ES (and ECDH-SS) when appiled to EC/EC2 keys is non-cofactor, which
> makes it unsuitable for curves with h > 1. X25519 and X448 have built-in
> workaround.
>
> Currently, there are no curves that trigger the problem. However, there
> are proposals to register curves that do.
>
> There are essentially two ways to address this:
>
> - Define copies of existing ECDH-ES/ECDH-SS codepoints that are limited
>   to EC/EC2 and multiply by h at output.
>

This sounds like your recommendation, can you elaborate a bit more on what
this might look like?


> - Change the existing codepoints to multiply by h at output for EC/EC2.
>
>
> The issue with latter is if somebody is using ECDH-ES/ECDH-SS with some
> private EC crv that happens to have h > 1. That would create an
> incompatibility.
>
>
>
>
>
> -Ilari
>
> _______________________________________________
> jose mailing list
> [email protected]
> https://www.ietf.org/mailman/listinfo/jose
>


-- 


ORIE STEELE
Chief Technology Officer
www.transmute.industries

<https://transmute.industries>
_______________________________________________
jose mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/jose

Reply via email to