cfrg/draft-irtf-cfrg-bls-signature

clarify that octets_to_point should be injective?

kwantam opened this issue ยท 1 comments

If octets_to_point isn't injective, there could be malleability-like issues (i.e., swap one byte-string for another that represents the same point). Should we require octets_to_point to be injective?

One can argue that this would be considered "naturally implied" requirement, but it's only appropriate to be explicit, so that nobody would be able to say "the requirement is not spelled, hence there is no such requirement." In other words ๐Ÿ‘ ๐Ÿ‘ ๐Ÿ‘ Oh! Did I say ๐Ÿ‘ ?