Re: [TLS] Simplify, simplify, simplify

> On May 14, 2014, at 22:40, Michael D'Errico <mike-list@pobox.com> wrote:
> 
> How do you get around the way version negotiation works?  If a client
> asks for TLS 1.3, and the server doesn't support it so it replies with
> TLS 1.2, the client has to know how to speak 1.2.

No, at that point the client can refuse to speak TLS 1.2 and abort the connection,

... then restart a new connection using it's legacy TLS 1.0 implementation. 

>  There's no way for
> the client to say it supports TLS 1.0 and 1.3, but not 1.1 or 1.2

You could also have out of band indications of what the servers support, so that you avoid the reconnection in most cases. 

I don't see a simple and efficient way though, so I guess you have a point...

> 
> Mike
> 
> 
> 
> Watson Ladd wrote:
>> Dear all,
>> As we are about to embark on a meeting to produce a protocol, let me
>> bring forwards what a veteran IETF participant believes about
>> protocols:
>> https://www.w3.org/2014/strint/papers/34.pdf
>> If TLS 1.3 is easier to deploy then TLS 1.2 it will be deployed.
>> Otherwise it will not be. Some of the specific recommendations are for
>> other WGs (in particular DANE, DNSSEC, and whoever is in charge of the
>> mess that is PKIX), but I think that the bottom-line message is worth
>> listening to: simplicity is the single most important criterion for a
>> security protocol.
>> Sincerely,
>> Watson Ladd
> 
> _______________________________________________
> TLS mailing list
> TLS@ietf.org
> https://www.ietf.org/mailman/listinfo/tls