Re: [AVTCORE] draft-ietf-avtcore-rfc5285-bis-05 -- fun special case

[Magnus Westerlund <[e-mail]>]

Hi,

If one follows RFC5285, then one can setup a number of formats with IDs 
in 1-14 range. Then one can decide on RTP stream level which format one 
or two byte. However, one you started sending one or two-byte in a 
particular RTP stream, you are not allowed to switch.

In addition the sender doesn't know if two-byte will work. It can assume 
that one-byte is available. So there is clearly issues here about 
capability determination from a sender perspective. Which is why RFC 
5285bis exists.

So the limiation on length is clearly present. I think the question is 
if this needs to be lifted in this part or not. I would note that 
actually the rule needs to be considered in the context of a particular 
RTP stream. Because one can actually consider that one RTP stream will 
only send a sub-set of header extension and thus be capable of using 
one-byte headers in this RTP stream, but another RTP stream requires the 
2-byte header and thus should use that from the start if the 
extmap-allow-mixed is not negotiated.

Looking at the -05 and -06 I can't find this rule in either version. 
Where was it included?

Based on that I can't find the context for this, I can't judge how 
important it is to make any changes.


Cheers

Magnus


Den 2017-02-02 kl. 04:09, skrev Dale R. Worley:
> Suppose that in the offer and the answer all of the identifiers are less
> than 15.  And suppose that mixing one-byte and two-byte formats is *not*
> negotiated.
>
> RTP packets are sent containing one-byte extension headers, and
> everything is going along well, and suddenly, the current data part for
> an extension is over 16 bytes.  That data can't be encoded in the
> one-byte format.
>
> I think there needs to be a rule that a UA can't commit to the one-byte
> format unless all of the extensions that might be sent are defined to
> have data lengths between 1 and 16 bytes.  So instead of the current
> rule
>
>     A client MUST NOT send an extension header with the one-byte format
>     if any negotiated valid identifier exceeds 14 and extmap-allow-mixed
>     is not negotiated.
>
> we need
>
>     A client MUST NOT send an extension header with the one-byte format
>     if:
>         (1) extmap-allow-mixed is not negotiated, and
> 	(2) (a) some negotiated valid identifier is more than 14, or
> 	    (b)some negotiated extension may have a data length of 0
> 	    bytes or more than 16 bytes.
>
> Dale
>
> _______________________________________________
> Audio/Video Transport Core Maintenance
> [e-mail]
> https://www.ietf.org/mailman/listinfo/avt
>


-- 

Magnus Westerlund

----------------------------------------------------------------------
Services, Media and Network features, Ericsson Research EAB/TXM
----------------------------------------------------------------------
Ericsson AB                 | Phone  [phone]
Färögatan 6                 | Mobile [phone]
SE-164 80 Stockholm, Sweden | mailto: [e-mail]
----------------------------------------------------------------------