W3C home > Mailing lists > Public > public-media-fragment@w3.org > December 2008

Re: Combining media fragments with other time-clipping methods

From: Yves Lafon <ylafon@w3.org>
Date: Wed, 10 Dec 2008 03:49:07 -0500 (EST)
To: Conrad Parker <conrad@metadecks.org>
cc: Media Fragment <public-media-fragment@w3.org>
Message-ID: <Pine.LNX.4.64.0812100334390.18974@ubzre.j3.bet>

On Fri, 5 Dec 2008, Conrad Parker wrote:

>> It can be a fragment, if there is a good reason for that. You can also (in
>> the case of pictures extracted) have metadata in the picture pointing to the
>> original video.
>> Now what will be more useful in this case, define it as a full class URI, or
>> as a fragment URI?
>
> I'm trying to understand this question :-)
>
> So, for example in a current implementation:
>
>  http://www.annodex.net/cmmlwiki/OSSForum-Trailer.png?t=0:02:10
>
> is an image representation of the first frame of the video subview:
>
>  http://www.annodex.net/cmmlwiki/OSSForum-Trailer.axv?t=0:02:10
>
> These are normal URLs; the first is used in <img>, the second in
> <video>. Both work with wget, and their relationship is clear. though
> application-specific. Here they have different use-cases. For a larger

Their relationship is indeed clear, there is no relationship between those 
two URIs.

> archive it would also make sense for these to be implemented on
> separate servers and thus appear as different resources, because they
> require different processing.

What matters is the external representation, so in that case, the image is 
another URI than the movie, so not a fragment. What happens server side 
(extraction at each request, internal cache of this process, generation of 
all images when the movie was uploaded, etc...) should be completely 
hidden to the clients.

> Now to the above question:
>
> As I understand it, the suggestion is to define a common URI
> mechanism, something like:
>
>  http://www.annodex.net/cmmlwiki/OSSForum-Trailer.axv?t=0:02:10#type=image/png
>
> or
>
>  http://www.annodex.net/cmmlwiki/OSSForum-Trailer.axv#t=0:02:10&type=image/png
>
> and to specify how every user agent should translate these into URLs
> without '#' (or a combination of URL + new request headers).

The server will receive everything before the "#",
http://www.annodex.net/cmmlwiki/OSSForum-Trailer.axv?t=0:02:10
and http://www.annodex.net/cmmlwiki/OSSForum-Trailer.axv are also 
unrelated URIs.

> If that is the suggestion, I suppose the benefit is to get rid of
> application-specific relationships and define a common syntax. Of
> course this adds a layer of indirection so we'd need to discuss the
> trade-offs.
>
> Do I understand the question correctly?

My question was more on the lines of "if you extract a picture from a 
video file", is it a fragment (ie: identified as such by using #), or is 
it another URI ? (and in that case we need to make the relationship with 
the video it was extracted from explicitely, like using
Link: rel="part_of" <video_uri>

-- 
Baroula que barouleras, au tiéu toujou t'entourneras.

         ~~Yves
Received on Wednesday, 10 December 2008 08:49:19 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Wednesday, 21 September 2011 12:13:31 GMT