Re: [bfcpbis] AUTH48 [JM]: RFC 8855 <draft-ietf-bfcpbis-rfc4582bis-16.txt> NOW AVAILABLE

[Christer Holmberg <[e-mail]>]

Hi,

...

>>>2) make the reference to RFC 5389/STUN an informative reference as is already done for RFC 5245/ICE
>>
>>My suggestion would actually be to update the ICE reference to RFC 8445. Again, that would align with other C238 cluster documents that reference ICE.
>
> [cue] Good point. This reference has already been updated to RFC 8445 as part of AUTH-48.

Good.
 
>> And, in RFC 8445 STUN is a *normative* reference.
>
> [cue] In rfc4582bis, the reference to ICE is currently Informative, whereas the reference to STUN is Normative. I believe they should either both be Normative or both be Informative. 
> Do you happen to know which way would be more consistent with similar references to STUN/ICE in cluster C238? In rfc4583bis, the reference to ICE is currently Normative.

Correct. rfc4583bis actually defines ICE procedures for BFCP, so Normative is fine.

4582bis contains the following sentence:

   "In order to facilitate the initial establishment of NAT bindings, and
   to maintain those bindings once established, BFCP entities using an
   unreliable transport are RECOMMENDED to use STUN [12] Binding
   Indication for keep-alives, as described for ICE [17]."

As you can see, it is actually described in the ICE spec on how to use STUN. So, therefore I agree that both should be either Normative or Informative. And, since there is a "RECOMMENDED", I assume that means they would have to be Normative.

Regards,

Christer



    On 5/21/20, 1:55 PM, "Keith Drage" <[e-mail]> wrote:

        I do note that the normative requirement is at SHOULD strength.

        However there are no statements that support an implementor to decide 
        under what conditions the requirement can be ignored, or the 
        consequences of ignoring, one or other of which should really be there.

        The lack of that information does not help in trying to evaluate the 
        consequences of updating the reference, versus leaving the reference as 
        it is - note that there is a risk both ways.

        The quick review I did, resulted in my feeling that the upgrade would be 
        OK, but I am not an expert in that area. Certainly I think either way it 
        should go to the WG for a quick review.

        Keith

        On 21/05/2020 19:09, Jean Mahoney wrote:
        > Hi Charles,
        >
        > On 5/21/20 11:31 AM, Charles Eckel (eckelcu) wrote:
        >> Hi Jean,
        >>
        >> I am more comfortable sticking with RFC 5389 at this point in time.
        >
        >
        > Ack.
        >
        >
        >> That said, I know that in another thread Christer made the point of 
        >> being consistent across the cluster in terms of referencing RFC 5389 
        >> vs. RFC 8489. If the decision is for the cluster switch to RFC 8489, 
        >> we can consider that. However, I think we would need to take it back 
        >> to the working group to see if there are any issues because I do not 
        >> believe the working group was not tracking this update and did not 
        >> anticipate its publication.
        >
        >
        > FWIW, so far, the one C238 document that has updated their STUN 
        > reference to RFC 8489 has an informative ref to it, not a normative one.
        >
        > Thanks!
        >
        > RFC Editor/jm
        >
        >
        >>
        >> Cheers,
        >> Charles
        >>
        >> On 5/20/20, 12:36 PM, "Jean Mahoney" <[e-mail]> wrote:
        >>
        >>      Authors,
        >>
        >>      We note that this document has a normative reference to RFC 5389 
        >> (STUN),
        >>      which was obsoleted just recently by RFC 8489.  Do you wish to 
        >> update
        >>      this reference to RFC 8489?
        >>
        >>      Thanks!
        >>
        >>      RFC Editor/jm
        >>
        >>

    _______________________________________________
    bfcpbis mailing list
    [e-mail]
    https://www.ietf.org/mailman/listinfo/bfcpbis