Hi all, At IETF 126 we discussed whether draft-ietf-jose-pq-composite-sigs should keep the raw, fixed-length ECDSA encodings introduced in -02/-03, or align with the DER-based encoding used by the LAMPS Composite document (draft-ietf-lamps-pq-composite-sigs) by using Ecdsa-Sig-Value (RFC3279) and ECPrivateKey (RFC5915).
A poll was organized, with the following result: "I prefer aligning with LAMPS Composite document?" yes: 21 / no: 12 / no opinion: 10 As a result, I updated our draft (not yet submitted to the Datatracker, but you can see the Editor's Copy on the Github repository: https://ietf-wg-jose.github.io/draft-ietf-jose-pq-composite-sigs/draft-ietf-jose-pq-composite-sigs.html#name-ecdsa-encodings) by introducing in particular a Section 4.5 "ECDSA Encodings". This section tries to explain in a simple way how to encode the ECDSA keys and signatures, by giving the bytes required so that no ASN.1 knowledge is needed, and highlights that implementers don't have to code a full ASN.1 parser. Test vectors have also been updated. The main arguments for aligning with LAMPS (by having DER for the ECDSA component) heard during the meeting were that: - it avoids maintaining two incompatible encodings of the same composite algorithm across X.509/CMS/TLS and JOSE/COSE. Libraries supporting both contexts would otherwise need two code paths; - the newly introduced ASN.1 is quite simple and doesn't require a complex ASN.1 parser. The Editor's Copy of the draft on Github tries to give everything which is needed for implementers to have the right encoding without ASN.1 knowledge; - we can think of those keys and signatures not as ASN.1 DER encoded, even if it's technically the case, but as opaque blobs. I know ASN.1 is not used in JOSE/COSE, and that whatever decision is made would disappoint someone, but aligning with LAMPS currently seems like the solution most people could live with. Is that correct? Any opinions on this matter, comments, or suggestions to update the section on ECDSA encodings are welcome! Thanks, Lucas
_______________________________________________ COSE mailing list -- [email protected] To unsubscribe send an email to [email protected]
