[TLS] Re: ML-KEM IANA and draft-connolly-tls-mlkem-key-agreement codepoint and inconsistencies

[Daniel Van Geest <[e-mail]>]

On 2025-03-06 1:05 p.m., Viktor Dukhovni wrote:

On Thu, Mar 06, 2025 at 11:29:16AM +0000, Daniel Van Geest wrote:



IANA went with the uppercase naming, while OpenSSL went with lower.



That is not the case, OpenSSL uses the IANA names, but the matching is
now case-insensitive.  Perhaps you were testing with the oqsprovider?

    $ openssl list -tls1_3 -tls-groups | tr ':' '\n' | tail -n6
    MLKEM512
    MLKEM768
    MLKEM1024
    SecP256r1MLKEM768
    X25519MLKEM768
    SecP384r1MLKEM1024


Apologies, I was scanning the code and came across the TLS_SIGALG_ENTRY for ML-DSA, which is lowercase, and too quickly added the case concern to my reference concern for ML-KEM.




Lowercase is generally more consistent with the other IANA codepoints,
with some exceptions including the recent hybrids. For what it's
worth, the slightly different bikeshed shade that I prefer is
lowercase.



We'll likely shortly change the group name lookup code to be
case-insensitive, just in case, and a convenience for the users.