[Openal-devel] Faster buffer model, and more random stuff

[Chris Robinson]

On Wednesday 27 February 2008 06:39:33 pm William Stewart wrote:
> I'm wondering about this for a couple of reasons.
>
> Isn't the fact that the developer is using ADPCM as their format
> implicitly saying that they care more about memory usage than CPU
> usage?

Not necessarilly, I think. They could be using ADPCM to keep their filesizes 
down, without jumping to something more heftier like Vorbis, and just rely on 
OpenAL to decode it so they don't have to. Just about all OpenAL 
implementations support it, so it's probably not asking much to rely on it.

But if it was something like Vorbis, yeah. The compression ratios you get with 
that bring up concerns (instead of 3.6:1, you're talking 10:1 or more).

> Following on from that - how much interest is there in providing
> support for this kind of functionality? If we added it to our
> implementation, would many of you use it?

Real-time ADPCM decoding? I've been giving it some thought. The question will 
be if alBufferDataStatic will be allowed to fail on specific formats, or if 
it has to support all the same formats alBufferData does. Each has their pros 
and cons..