On Fri, Sep 01, 2023 at 12:17:20PM -0500, Orie Steele wrote:
> On Fri, Sep 1, 2023 at 11:52 AM Ilari Liusvaara <[email protected]>
> wrote:
> 
> > On Fri, Sep 01, 2023 at 11:13:01AM -0500, Orie Steele wrote:
> > > On Fri, Sep 1, 2023 at 10:59 AM Ilari Liusvaara <
> > [email protected]>
> > > wrote:
> >
> > I don't see room for alignment.
> >
> > What would work for one will not work for any others, because the
> > signature frameworks are just different.
> >
> > E.g., COSE has both body and signature headers, JOSE has only signature
> > headers, and PKIX (LAMPS) does not have explicit headers.
> >
> > This makes it necressary to apply different approaches to all three.
> >
> >
> I think we are possibly agreeing, but speaking past each other.
> 
> I am asserting that the "raw crypto layer" needs alignment / can be
> aligned... the envelopes are of course different.

Does "raw crypto layer" mean raw cryptographic algorithms like
SHA256/SHA384/SHA512/ML-DSA-44/ML-DSA-65/ML-DSA-87, or something
else?

If the first, of course those will be common to all.


> If we added a named alg for "hash with shake256 then sign with dilithium
> 5"...

Beware that is prone to undesirable combinatorics. And might require
nasty hacks in other to not violate JWS/COSE_Sign requirements.

It would be a new parameter "hash with shake256" and alg "sign with
ML-DSA-87(sic)". 

Bit similarly as JOSE has alg "derive key with ECDH-ES" and enc "encrypt
with AES-128-GCM".

 
> Are you asserting that it would be "implemented differently" for COSE /
> JOSE / LAMPs ?

The differences I was talking are:

- What exactly is fed to the hash function.
- What exactly is fed to the signining function.
- How pre-hashing is denoted in the message. 

The last two are nothing new: All three feed different things to the
signing function and encode their messages differently.




-Ilari

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

Reply via email to