Hi,

- Looking through the COSE Algorithms registry, I find that there are three 
different models for ECC signatures:

1. ES256 = ECDSA + SHA-256
   ES384 = ECDSA + SHA-384
   ES512 = ECDSA + SHA-512
   Curve specified elsewhere.

2. EdDSA = EdDSA
   Curve specified elsewhere
   Hash algortithm determined by curve.

3. ES256K = ECDSA + SHA-256 + secp256k1
   Curve and hash algorithm included.

Is there any reason why ES256K was specified like this? My understanding is 
that 1. and 2. follow PKIX but not 3. My understanding is that that ECDSA + 
SHA-256 + secp256k1 in PKIX would be specified in the same way as ECDSA + 
SHA-256 + secp256r1

id-ecPublicKey + secp256k1
ecdsa-with-SHA256


- The COSE registries allow registration labels with a 1, 2, or 3 byte CBOR 
encoding. The registry differentiates between 2 and 3 byte encodings, but the 
registry does not differentiate between 1 and 2 byte encodings. Integers in the 
range [-24,23] have a one byte encoding.

A lot of a algorithms not suitable for constrained IoT such as HMAC 512/512, 
A256GCM, A256KW, direct+HKDF-SHA-512, etc. have been given 1 byte identifiers. 
These should maybe have been saved for algorithms that constrained IoT will 
likely use.

ECDSA25519 is targeting constrained IoT and should probably be given a 1 byte 
label, which is can't get unless the draft is changed to standards track.


Cheers,
John

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

Reply via email to