> > There are uses cases where multiple values of hkc::kem make sense (key > compatible with multiple KEMs, especially P-* vs CP-*). But HPKE-KEM key > has exactly one primary KEM. This is as hkc is about capability.
Even when a key can be used with multiple KEMs, I think it is reasonable to limit its usage to only a single KEM. I believe that hkc::kem should not be allowed to take multiple values. And making common key paramter that depends on kty is unprecedented, and > seems like something that would cause severe implementation problems. I don't think this is a convincing reason, but I understood that your proposal was based on this concern. In any case, it seems that our opinions are basically in alignment and we can resolve the subtle differences in our opinions through further discussion in the future. AJITOMI Daisuke 2023年5月2日(火) 18:06 Ilari Liusvaara <[email protected]>: > On Tue, May 02, 2023 at 04:10:01PM +0900, AJITOMI Daisuke wrote: > > > Using hkc there is a bad idea (it says what KEM is supported, not what > > KEM it is). > > > > I don't think it's a bad idea. > > In the new kty "HPKE-KEM", it may be possible to make "hkc" be MUST, so I > > think there is no need to hesitate to use the hkc::kem. > > There are uses cases where multiple values of hkc::kem make sense (key > compatible with multiple KEMs, especially P-* vs CP-*). But HPKE-KEM key > has exactly one primary KEM. This is as hkc is about capability. > > And there may be situations where it is possible to convert key from one > KEM to another (e.g., P-* to CP-*). > > And making common key paramter that depends on kty is unprecedented, and > seems like something that would cause severe implementation problems. I > remember common header parameter depending on alg being shot down. > > > > -Ilari > > _______________________________________________ > COSE mailing list > [email protected] > https://www.ietf.org/mailman/listinfo/cose >
_______________________________________________ COSE mailing list [email protected] https://www.ietf.org/mailman/listinfo/cose
