[ogg-dev] OggPCM2 : chunked vs interleaved data

[Jean-Marc Valin]

Le mardi 15 novembre 2005 à 19:58 +1100, Erik de Castro Lopo a écrit :
> The remaining issue to be decided for the OggPCM2 spec is the support 
> of chunked vs interleaved data.

My take on this is that we definitely need to support interleaved format
as this is the format used by WAVE, most soundcard APIs and just about
everything.

I'm not as sure about the other one, but I think we should not support
chunked format because:
1) It makes the implementation more complicated
2) Few applications use it so it not too hard to convert for these
3) Even if we support chunked, apps that want chunk data would still
need to have code to support interleaved.

That being said, I might change my mind on chunked support if you think
I overlooked something.

	Jean-Marc

> Just so that everyone understands what we are talking about, consider a
> stereo file that gets stored as an OggPCM file. Within an OggPCM packet,
> the audio samples for the left and right channels can be stored as 
> interleaved where the samples would be:
> 
>     l0, r0, l1, r1, ..... lN, rN
> 
> or as chunked where the samples would be:
> 
>     l0, l1, l2, ..... lN, r0, r1, r2, .... rN
> 
> 
> So the question is do we support:
> 
>     a) interleaved
>     b) chunked
>     c) both
> 
> This is file level storage. An API for reading this data could still
> allow the demulitplexing/re-multiplexing on the fly.
> 
> Opinions?
> 
> Erik