[theora] Indexing Ogg files for faster seeking

[Chris Pearce]

On 10/14/2009 3:21 PM, Conrad Parker wrote:
> Players should at least use the presentation time, ie. not display
> earlier frames.
>
> The BOS page you have specified is mostly the same as the BOS page for
> Skeleton anyway. I think it'd make more sense to rev Skeleton to
> include a presentation end-time than to introduce a whole new BOS
> page.
>
> Presentation end-time would also be useful for players when the file
> is a chopped section, to say that after this point you may not have
> all required audio+video.

The skeleton's "presentation time" isn't the presentation time of the 
first frame in the media? Just the time at which you want to start 
playing? I didn't realise that. So it's valid to have the "presentation 
time" set to be half way through the physical media, and by doing so 
ensure that a valid decoder will download, decode and discard half the 
media? Now why would you want to do that?

Chopping media I can understand, but why ship data which you're not 
decoding? That seems silly to me.

Why would I want to include the index in the skeleton? That would 
require players to implement its more controversial features such as 
this in order to get the benefit of the index.

The index should index the entire media, not the "presentation window" 
as specified by the skeleton track, as if a player chooses to ignore the 
skeleton's presentation window we still want the index to be as useful 
as possible.

The index belongs in its own track. Best to not confuse the skeleton's 
presentation window with the actual medias boundaries.

Chris P.