Re: [TLS] WG last call of draft-ietf-avtcore-rfc5764-mux-fixes-05

[Joseph Salowey <[e-mail]>]

On Wed, Mar 2, 2016 at 2:22 PM, Ilari Liusvaara <[e-mail]>
wrote:

> On Wed, Mar 02, 2016 at 02:08:28PM -0800, Joseph Salowey wrote:
> > Reserving large portions of other protocols number spaces is not a good
> way
> > to do things.   This will quickly become unworkable if other protocols
> > decide to do the same thing.  This type of behavior needs to be
> > discouraged.  There is no guarantee that the multiplexing scheme
> prompting
> > this registration request will work with TLS 1.3 or any future version of
> > TLS.
>
> Well, outside of some very exotic extensions (none are defined to date,
> and I'm having problems even imagining what such things would actually
> do), it would work in TLS 1.3 Editor's Draft (since it only ever uses
> 21, 22 and 23[1]).
>
> Future versions of TLS are another matter...
>
> And even if there was some extension to omit the version from record
> header, it would still work (the demuxer would however need to be
> aware of that extension).
>
>
> [1] For comparision, TLS 1.2 uses 21-23 and also 20 and in presence of
> certain (infamous) extension, 24.
>
>
>
[Joe] Yes you are correct that the current draft includes a dummy header,
 however the working group has discussed not using the dummy header which I
believe would cause problems for the multiplexing scheme because the
location where the contentType was would no longer contain the contentType.
   It seems unlikely that we would make this change at this point, however
there are some folks who would like to get rid of the dummy header.



> -Ilari
>