Leaving the registry for later sounds fine to me, in particular if it isn’t 
clear yet what the registration policy should be.

If we ever add not-over-IP “transports”, there might be a need for experiments 
before we go registering.
So I would probably identify a range of numbers that are strictly reserved for 
use in experiments.
(This needn’t be very small; say, 65280 to 65535.)

Grüße, Carsten


> On May 29, 2017, at 04:51, Brian E Carpenter <[email protected]> 
> wrote:
> 
> On 23/05/2017 10:40, Adam Roach wrote:
> ...
>> The CBOR definition has constants for IP_PROTO_TCP and IP_PROTO_UDP, but
>> no way to register additional values with IANA. This does not seem
>> future-proof.
> 
> Adam is correct. The current values (6 and 17) are of course values from
> https://www.iana.org/assignments/protocol-numbers/protocol-numbers.xhtml
> and the names are those used in the socket API. If we wanted to add, say,
> SCTP it would be easy: IP_PROTO_SCTP = 132.
> 
> The problem comes if we want to add "transport" protocols that aren't
> directly over IP: things like HTTP, COAP, QUIC... for example. There's
> no registry for them.
> 
> We have considerable flexibility thanks to CBOR; for example, as Michael
> Richardson noted, we could define values >255 for transport protocols that
> are *not* directly over IP. However, that would need a new IANA registry.
> 
> Proposal: Note in the text that the current values are taken from the
> existing Protocol Numbers registry. Also note that if values are required
> in future that are not in that registry, a new registry for values >255
> will be created. So IANA doesn't have to do anything now.
> 
> Opinions? Objections?
> 
>    Brian
> 
> _______________________________________________
> Anima mailing list
> [email protected]
> https://www.ietf.org/mailman/listinfo/anima
> 

_______________________________________________
Anima mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/anima

Reply via email to