[openpgp] Re: I-D Action: draft-ietf-openpgp-persistent-sy mmetric-keys-02.txt

On Wednesday, November 5th, 2025 at 11:54, Andrew Gallagher <[email protected]> wrote:

> Sure, but the ink is barely dry on AEAD modes, and mapping 0 explicitly
> to AEAD assumes that all future extensions will fit in the AEAD
> paradigm. What if in 2035 we decide there's a better construction than
> AEAD, just like we decided that AEAD was better than MDC?

The AEAD paradigm/API is quite flexible, we already have a few
completely different constructions that fit in it (AES-GCM vs
ChaChaPoly, for example). If we ever come up with something that
requires a completely different API, I think defining a new packet
version could also be warranted (as we'd also need an SEIPDv3).

> Stepping back a bit, the prior art in the other registries is that "0"
> means "None", and if we can construct something future-proof using this
> interpretation, let's do it. But if we need to put an asterisk beside
> "None" to say "(* None means AEAD)" then I think we're stretching, and
> should maybe just use a normal code point?

Personally I think that in the public key algorithms registry,
saying 0 = no public key algorithm, use symmetric crypto instead,
makes sense. How we want to spell that in the registry is of course
up for debate, but I think it also makes sense that symmetric crypto
means AEAD.

Having 0 point to a registry with a single entry ("AEAD") which in
turn points to another registry (AEAD algorithms) seems like overkill
that we'll never need, IMHO.

Best,
Daniel

_______________________________________________
openpgp mailing list -- [email protected]
To unsubscribe send an email to [email protected]