[openpgp] Re: session key length with SEIPDv2

[Daniel Kahn Gillmor <[e-mail]>]

On Sat 2024-10-05 15:42:19 +0000, Daniel Huigens wrote:
> All of that is not to say that I'm opposed to consumer guidance, but I
> think (normative) producer guidance is the more important part and the
> part that can more unambiguously be shown to have been intended given
> the above quotes of RFC9580, and thus I think it could go in an erratum?
> Whereas if we want consumer guidance (for SEIPDv2 and v1) I'd expect
> it'd have to go somewhere else. But, lmk if you disagree, of course.

Producer guidance without consumer guidance seems rather toothless to
me, but i'd be curious to know what the rest of the WG thinks about
this.

Here's a stab at fairly wishy-washy proposed guidance for both sides
that tries to explicitly state what i expect is the status quo:

     When producing a message using SEIPD v2 with corresponding PKESKs
     or SKESKs, The length of the session key MUST equal the key size of
     the symmetric algorithm used in the encryption container.

     When handling a message that uses SEIPD v2, if the session key size
     does not equal the key size of the symmetric algorithm used in the
     encryption container, the consuming implementation SHOULD warn that
     the message is malformed, and MAY decline to decrypt the message.

This consumer guidance is itself still fairly toothless, but at least it
describes the range of likely responses that justify the MUST on
interoperability grounds.

wdyt?

        --dkg