Re: [TLS] Updated EdDSA in TLS drafts

[Simon Josefsson <[e-mail]>]

> >> My v. high level feedback is that we really don't want
> >> to get too far ahead of cfrg here. They've not yet got
> >> all details of their preferred signature scheme using
> >> that curve defined and it'd be really dumb for any IETF
> >> WG to get ahead of that by accident.
> >>
> >> So while working on this draft is a fine thing, I don't
> >> think we want to get to IETF LC for things like this until
> >> after cfrg have figured out their preferred signature
> >> scheme. (And the cfrg chairs tell me that ought be done
> >> soon - in a few months.)
> > 
> > Thanks for feedback.  I believe these documents are quite far away
> > from LC-ready so that this aspect shouldn't be a problem.
> 
> Yep, I think that's correct. But I'd be somewhat wary of
> making design decisions in the meantime - we really don't
> want to end up with a cfrg-annointed signature scheme that
> ends up being hard to integrate because of decisions made
> now. (And folks who care about what cfrg are doing on that
> should get involved on the cfrg list.)

Hi Stephen.  Sure.  I don't think any decisions are or will be made
here now.  This is brainstorming about how to approach adding a new
signature algorithm to TLS.  There are no changes to any TLS version
proposed that could cause integration problems for a future CFRG-blessed
signature scheme.

> > I view the EdDSA/Ed25519 effort similar to the Brainpool/GOST
> > efforts. If CFRG happens to pick EdDSA, then fine, but that isn't
> > the motivational factor here.
> 
> Well, maybe. With no hats, I personally don't think we want
> more unused ciphersuites, and especially not ones that are
> almost-but-not-quite the same as one that uses the eventual
> cfrg-annointed signature scheme. (Unless there's a good
> reason to not use the cfrg stuff of course, in which case
> we're in a different situation.)
> 
> But so long as the TLS WG don't bake in too much stuff and
> so long as cfrg continue to make progress in a timely fashion
> it should all work out.

Agreed.  Btw, it seems that no new ciphersuites will have to be
allocated anyway, and that the existing ECDSA ciphersuites can be
reused.

/Simon