Re: [TLS] Inconsistent extension definitions

On Sun, Jun 23, 2019 at 7:51 PM Yishuai Li <yishuai@upenn.edu> wrote:

> Dear TLS working group,
>
> Here’s a duplicate of GitHub issue tlswg/tls13-rfc#21 I opened today,
> which somehow disappeared:
>

I closed it. The RFC has been published, so filing issues on that repo
isn't useful.

Supported Versions are defined as Variants:
>
>     struct {
>         select (Handshake.msg_type) {
>             case client_hello:
>                  ProtocolVersion versions<2..254>;
>
>             case server_hello: /* and HelloRetryRequest */
>                  ProtocolVersion selected_version;
>         };
>     } SupportedVersions;
>
> while Key Share is defined as separate Constructed Types:
>
>     struct {
>         KeyShareEntry client_shares<0..2^16-1>;
>     } KeyShareClientHello;
>
>     struct {
>         NamedGroup selected_group;
>     } KeyShareHelloRetryRequest;
>
>     struct {
>         KeyShareEntry server_share;
>     } KeyShareServerHello;
>
> Is there a specific reason for choosing different definition styles?
>

I'm not aware of any specific reason. I think it's just an inconsistency,
perhaps due to aesthetics.

The protocol is the same eithere way.

Is it worth unifying them?
>

Probably not. It's editorial and there is no real mechanism for doing that,
given that the RFC has been published.

-Ekr

> Also, is this mailing list the right place for such questions?
>
> Thanks,
> Yishuai Li
>
> _______________________________________________
> TLS mailing list
> TLS@ietf.org
> https://www.ietf.org/mailman/listinfo/tls
>