[rtcweb] Fwd: inconsistency in drafts: rtcp-fb and rtcp-rsize

Rachel Dvori <rachel.dvori@gmail.com> Mon, 27 January 2014 15:53 UTC

MIME-Version: 1.0
In-Reply-To: <CANCLmwnfxNoDB-xpf=QDeYhJYkak8qmWpFHBSZaFsWN2gWnOpg@mail.gmail.com>
References: <CANCLmwnfxNoDB-xpf=QDeYhJYkak8qmWpFHBSZaFsWN2gWnOpg@mail.gmail.com>
Date: Mon, 27 Jan 2014 17:52:55 +0200
Message-ID: <CANCLmwktXP=Mj+BHGdq+vO=WCmKD6mi1rcHVruhHenxfo2GX2Q@mail.gmail.com>
From: Rachel Dvori <rachel.dvori@gmail.com>
To: rtcweb@ietf.org
Content-Type: multipart/alternative; boundary="f46d0442836a7eecaf04f0f5b31b"
Subject: [rtcweb] Fwd: inconsistency in drafts: rtcp-fb and rtcp-rsize
Precedence: list

The basic question is:

-          In WebRTC offer, do we *always* have to include both attributes
of   rtcp-fb and rtcp-rsize ?  (even if we do not have any feedback msg to
send ?)







When looking into the draft of :
rtcweb-rtp-usage<http://tools.ietf.org/html/draft-ietf-rtcweb-rtp-usage-11>

It is indicated that both below are MANDATORY (MUST).



Regarding *Reduced Size RTCP:*

 Support for non- compound RTCP feedback packets
[RFC5506<http://tools.ietf.org/html/rfc5506>]
is REQUIRED , but MUST be negotiated using the signalling channel before
use.



Regarding *RTCP feedback:*



For WebRTC use, the Extended  Secure RTP Profile for RTCP-Based Feedback
(RTP/SAVPF) [RFC5124 <http://tools.ietf.org/html/rfc5124>], MUST be
implemented.  ….

….This builds on the basic …… the RTP profile for RTCP-based feedback
   (RTP/AVPF) [RFC4585 <http://tools.ietf.org/html/rfc4585>], ….



>From above, one can conclude that the SDP of an offer *MUST* include *both*:

*-          **a=  rtcp-fb*

*-          **a = rtcp-rsize    *







%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

When looking into:
draft-ietf-rtcweb-jsep-05<http://tools.ietf.org/html/draft-ietf-rtcweb-jsep-05>
   (page 24)



It can be understood that only rtcp-rsize is MUST to be included in the
offer (and  rtcp-fb is not mandatory to be included in the offer)



Below is quote from that dratft:

-          For *each supported* RTCP feedback mechanism, an "a=rtcp-fb"…
  (My Remark:   so if there is no support, it is not mandatory to add it)

-          An "a=rtcp-rsize" line, as specified in [RFC5506], Section
5 <http://tools.ietf.org/html/rfc5506#section-5>.



   R-11  [RFC4585 <http://tools.ietf.org/html/rfc4585>] MUST be
implemented to signal RTCP based feedback.

   R-13  [RFC5506 <http://tools.ietf.org/html/rfc5506>] MUST be
implemented to signal reduced-size RTCP  messages.





%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%



And last draft: draft-nandakumar-rtcweb-sdp-03
<http://tools.ietf.org/html/draft-nandakumar-rtcweb-sdp-03>



It indicates that:

A typical … communication session can be characterized as below:

-          Provides RTCP based feedback mechanisms,



It does not mention that rtcp-rsize is MUST and it is does not appear
in all examples of SDP’s





So there is inconsistency between the drafts.

I would appreciate VERY Much your comment on the first question:

 In WebRTC offer, do we *always* have to include both attributes of
rtcp-fb and rtcp-rsize ?  (even if we do not have any feedback msg to send
?)



And if not –  can anyone provide clarification on what is expected to be on
each SDP offer (even if we do not have fb msg to send, but it can be easily
added in a SDP, if needed)



With appreciation

RD

Re: [rtcweb] Fwd: inconsistency in drafts: rtcp-f… Harald Alvestrand
[rtcweb] Fwd: inconsistency in drafts: rtcp-fb an… Rachel Dvori