W3C home > Mailing lists > Public > public-media-fragment@w3.org > May 2010

Re: Notify user agent available fragment

From: Jack Jansen <Jack.Jansen@cwi.nl>
Date: Sun, 2 May 2010 22:59:36 +0200
Cc: public-media-fragment@w3.org
Message-Id: <AD34875F-53F8-4C8B-8651-9DA4537F3A80@cwi.nl>
To: Jeroen Wijering <jeroen@longtailvideo.com>

On  1-May-2010, at 11:04 , Jeroen Wijering wrote:

> Hello workgroup,
> 
> Pardon my ignorance if this question was already addressed. There seems to be no reference in the working draft, but I realize you could consider this topic out of scope.
> 
> How do you see the user-agent being notified in advance of the fragmentation possibilities of a certain resource?
> 
> For example, one video on a server might contain an AUDIO track (and thus the possibility to retrieve audio fragments), while the other video might not contain this track. As a second example, that first video might have a VIDEO track with keyframe-interval of two seconds and the second video might have a VIDEO track with keyframe-interval of four seconds. The server might not be able to return a 2-second fragment of the second video without transcoding (which should be avoided).
> 
> User agents would like to have this exact information in advance, without relying on fallback actions (e.g. the server returning an error in the first example and a 4-second fragment in the second example). They would be able to present the appropriate interface and apply the appropriate resource fetching heuristics.


I think we consider it out of scope. But: note the slight uncertainty implicit in the "I think" bit of the sentence:-), we'll see what other people say.

What we definitely consider to be out of scope is discovery. While we think it would be good if there was some scheme through which an interested client could get information on a media resource (duration, track names, size, etc) we don't want to incorporate this in the first version of Media Fragments. One reason is that it relies on work being done by the Media Annotations WG. Another is that we want to standardise only the things we think we understand well enough, and this isn't one of them. What you want (an answer to the question "Can I ask for track#2", if I interpret your message correctly) is pretty close to discovery.

There is also a practical consideration: if a user expects a user agent to handle <http://example.com/example.ogg#t=10,20> the user will most likely also expect <file://home/example.ogg#t=10,20> to be handled. That means that a user agent must always be ready to do the implementation without server help. 
--
Jack Jansen, <Jack.Jansen@cwi.nl>, http://www.cwi.nl/~jack
If I can't dance I don't want to be part of your revolution -- Emma Goldman
Received on Sunday, 2 May 2010 21:00:12 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Wednesday, 21 September 2011 12:13:38 GMT