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

[Eric Rescorla <[e-mail]>]

On Sat, Jul 1, 2017 at 7:56 PM, John Levine <[e-mail]> wrote:

> In article <CABcZeBOs1yZ7q3oBgNeVkw=zSQb_SuS4hqK8BH0ebrD5LRYTFg@mail.
> gmail.com> you write:
> >In any case,  it's not true that eddsa keys are 256 bits. Eddsa keys are
> >256 bits when you are using X25519 but not when you are using X448. To the
> >extent to which you believe that it's redundant to have both non-FP and FP
> >variants, the answer is to *always* use a fingerprint.
>
> The only eddsa algorithm we're adding is ed25519.
>

Right now yes, but I'm interested in thinking past this document.


The point of the fingerprints was to make it easier to use long RSA keys.


That may have been your point, but when I suggested it, my objective was to
future proof the protocol against other algorithms with larger keys as well.



>   We're
> stuck with unfingerprinted RSA keys because they're what we have now.
>

Yes, but that doesn't mean we need to continue that trend for future
algorithms.

-Ekr


>
> R's,
> John
>