Re: [Dcrup] I-D Action: draft-ietf-dcrup-dkim-crypto-03.txt

["John R Levine" <[e-mail]>]

> Given that (a) we already have algorithms that are bigger and (b) 
> post-quantum keys might be quite a bit bigger, it seems unwise to design 
> under the assumption that it will not happen.

Well, OK, but concretely, what do you want us to do now?

DKIM has had algorithm tags since it was defined a decade ago.  This draft 
adds two new algorithm tags (rsafp-sha256 and eddsa-sha256) and deprecates 
one old one (rsa-sha1) and as part of that adds RSA key hashes as a key 
verification method.  If it turns out that we need to switch signature 
algorithms, we'll revise DKIM again, and add and deprecate tags and 
algorithms as needed.  Maybe new algorithms will use key hashes, maybe 
they won't, but at that point we can define whatever we need.

Right now, today, I don't see that hasned ed25519 keys solve any problem, 
so I'm not going to include them unless there is something I don't 
understand that they can do that plain ed25519 keys in the key records 
don't.

R's,
John