W3C home > Mailing lists > Public > public-web-and-tv@w3.org > December 2011

Re: [MPTF] ADR_Min_Control and manifest control

From: Mark Watson <watsonm@netflix.com>
Date: Fri, 16 Dec 2011 18:12:09 +0000
To: Jan Lindquist <jan.lindquist@ericsson.com>
CC: "public-web-and-tv@w3.org" <public-web-and-tv@w3.org>
Message-ID: <B2F6DE0E-E3FA-4BAD-A2FA-FAF2DFAD1A58@netflix.com>

On Dec 15, 2011, at 11:48 PM, Jan Lindquist wrote:


I appreciate the references that were added relating to Flash and Silverlight which adds clarity to why several of the parameters that were in the draft coming into yesterday's phone conference. The below comments tries to have create a more consistent draft.

I am still not convinced that we cannot have a control of the manifest. There isn't precedence with having this additional control of the manifest and selection of the tracks or levels. There is obviously an exposure of the manifest through error codes, representationID and callback functions reporting changes in the selected representation. Why are we so keen to exclude the control of the levels directly related to the manifest?

This would be "Model 2", with control of the adaptive streaming in Javascript. It is certainly viable, but I don't think we are anywhere close to having a proposal that makes sense. In other standards bodies I see detailed multi-page contributions with analysis and experimental results for far simpler topics than control of adaptive streaming. Here we have a few notes on a Wiki. We don't even have an initial consensus on the functional split between UA and Javascript. We need to be explicit about these things, not back into them from a handful of API proposals.

I still maintain that "Model 3" is the more appropriate route in the short term for Javascript control of adaptive streaming. We can be more confident about the functional split needed here because we are only asking the UA to provide a set of discrete low-level component capabilities (download through XmlHttpRequest, playback through webkitSourceAppend, or equivalents). Experimentation and deployment experience with this route is what will lead is to good designs for a Model 2 approach, if that is needed.


The use case CT2 specifically gives a clear example of the control there needs to be in place in order to get HD. It is not a bandwidth control that sets HD, it is the control of the manifest and quality level. There may be issues with associating a level to the different adaptive streams but that can be done by the respective adaptive stream standards just like texttracks to DVB mappings.

The CT3 will not be achieved through a minimumBandwidth either as explained in a seperate e-mail.

For support of the CT2 and CT3 I recommend to re-introduce the maxLevel and minLevel.

Received on Friday, 16 December 2011 18:12:39 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 17:57:11 UTC