Re: [rtcweb] draft-ietf-rtcweb-jsep-08

Paul Kyzivat <pkyzivat@alum.mit.edu> Wed, 05 November 2014 02:28 UTC

Message-ID: <54598B3E.3040207@alum.mit.edu>
Date: Tue, 04 Nov 2014 21:28:14 -0500
From: Paul Kyzivat <pkyzivat@alum.mit.edu>
User-Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.7; rv:24.0) Gecko/20100101 Thunderbird/24.6.0
MIME-Version: 1.0
To: Justin Uberti <juberti@google.com>
References: <54594A9B.9090703@alum.mit.edu> <CAOJ7v-3zgw1BByRm+AY74_AfD1DbiTKoR2YJ_7ojotY5D8XjHw@mail.gmail.com>
In-Reply-To: <CAOJ7v-3zgw1BByRm+AY74_AfD1DbiTKoR2YJ_7ojotY5D8XjHw@mail.gmail.com>
Content-Type: text/plain; charset="UTF-8"; format="flowed"
Content-Transfer-Encoding: 7bit
Archived-At: http://mailarchive.ietf.org/arch/msg/rtcweb/6N3D0si6zJ7HchixwzrqZal3jUQ
Cc: "rtcweb@ietf.org" <rtcweb@ietf.org>
Subject: Re: [rtcweb] draft-ietf-rtcweb-jsep-08
Precedence: list

On 11/4/14 7:58 PM, Justin Uberti wrote:
>
>
> On Tue, Nov 4, 2014 at 1:52 PM, Paul Kyzivat <pkyzivat@alum.mit.edu
> <mailto:pkyzivat@alum.mit.edu>> wrote:
>
>     I have a few related comments on jsep-08:
>
>     Section 4.1.1 defines the BUNDLE policy values. It certainly appears
>     that when you are using the default "balanced" policy there will be
>     *separate* bundle groups for audio and video, and that the SCTP
>     m-line will not be bundled.
>
>
> This is not correct. Rather, the m= lines for these groups will not be
> marked as a=bundle-only.

Well, it is your draft, so you should know what you intended for it to 
say. But I didn't get that from the words:

    balanced:  The application will BUNDLE all media streams *of the same
       type* together.  That is, if there are multiple audio and multiple
       video MediaStreamTracks attached to a PeerConnection, all but the
       first audio and video tracks will be marked as bundle-only, and
       candidates will only be gathered for N media streams, where N is
       the number of distinct media types.  When talking to a non-BUNDLE-
       aware endpoint, only the non-bundle-only streams will be
       negotiated.  This policy balances desire to multiplex with the
       need to ensure basic audio and video still works in legacy cases.
       Data channels will be in a separate bundle group.

    max-bundle:  The application will BUNDLE *all of its media streams,
       including data channels, on a single transport*. All streams other
       than the first will be marked as bundle-only.  This policy aims to
       minimize candidate gathering and maximize multiplexing, at the
       cost of less compatibility with legacy endpoints.

The presence of *all media streams of the same type* and *all of its 
media streams* suggests that those are key distinctions in those two 
definitions. If that wasn't the intent, then I suggest revising the 
above words to be more in line with your intent. Perhaps:

    balanced:  The application will BUNDLE all media streams
       together. If there are multiple MediaStreamTracks attached to a
       PeerConnection, candidates will be gathered for the first of
       each distinct media type, and the others will be marked as
       bundle-only. When talking to a non-BUNDLE-aware endpoint, only
       the non-bundle-only streams will be negotiated.  This policy
       balances desire to multiplex with the need to ensure basic audio
       and video still works in legacy cases. Data channels will be in a
       separate bundle group.

	Thanks,
	Paul

[rtcweb] draft-ietf-rtcweb-jsep-08 Paul Kyzivat
Re: [rtcweb] draft-ietf-rtcweb-jsep-08 Justin Uberti
Re: [rtcweb] draft-ietf-rtcweb-jsep-08 Paul Kyzivat
Re: [rtcweb] draft-ietf-rtcweb-jsep-08 Justin Uberti