Re: [TLS] WGLC for draft-ietf-tls-rfc4492bis

[Ilari Liusvaara <[e-mail]>]

On Wed, Dec 07, 2016 at 07:18:56AM +0900, Martin Thomson wrote:
> On 7 December 2016 at 03:24, Sean Turner <[e-mail]> wrote:
> > Just a reminder that this WGLC will close on Friday December 9th.
> 
> A timely reminder :)
> 
> I reviewed the document and it looks pretty good.  I'd have sent a PR
> with some minor changes to grammar.
> 
> The question I wanted to ask was how we wanted to manage the
> relationship with TLS 1.3, particularly for EdDSA.
> 
> The draft asks for a NEW codepoint in the hash and signature
> algorithms structure.  That clobbers a whole bunch of space that TLS
> 1.3 is going to rework.  I don't think it's a good idea to perform
> concurrent surgery on this registry, particularly since new codepoints
> have the effect of taking out new swathes of space.  At best we send
> confusing signals to IANA.
> 
> I would prefer to take the arrangement that we have in TLS 1.3 and
> backport it here so that we have a consistent story.  I also think
> that taking a single 2 octet codepoint from the SignatureScheme space
> is better all around.

I actually reviewed the document and noticed the exact same thing.


Also, in my TLS implementation, doing EdDSA in TLS 1.2 by backporting
the TLS 1.3 mechanism (using ECDSA legacy type) just fell naturally
out of the implementation. It was easier to have all the machinery
needed to handle it than not to have that machinery.


-Ilari