Re: [linux-audio-dev] CSL-0.1.2 Release

[Abramo Bagnara <abramo alsa-project org>]

Paul Davis wrote:
> 
> >Although I agree that many application don't need/want to cope with
> >hardware/software setting this is not a good reason to remove this
> >capability from API.
> 
> If the API you're talking about is ALSA, then I totally agree with
> you. ALSA *is* a HAL, and its appropriate for it to contain these
> aspects in its design.

ALSA is much more than a HAL as you already know: it's an API versatile
enough to talk *also* with hardware.

This is exactly the key concept I'd like you note: "versatile enough".

I insist that a right API need to have methods to get/set/query
properties for underlying objects.

> That doesn't mean that a HAL is the right API for applications that
> want to play or record audio data. I believe there is evidence (from
> other systems and introspection (!)) that the right level for
> applications is "above" a HAL, not a wrapper around it.
> 
> >A make another example: very few application classes need snd_pcm_rewind
> >but to remove it would be a great mistake. Games *need* it to do a good
> >job (as I've verified with John Carmack and others) then to remove it is
> >simply wrong.
> 
> Thats not true. DirectX provides no equivalent method to
> snd_pcm_rewind(), and neither does IRIX, nor does Mac OS X, and all
> these systems do/have/will support games. If John believes this, its
> because he doesn't understand how to do without it in the context of
> the applications he writes. That doesn't mean its necessary, it means
> that game writers have designed their application architectures around
> the existence of such a function.

No, they don't. They asked to me to make this feasible. The point is the
ability to send sample in advance and then retract them if it's needed
(by example because another event occurred). This help to reduce
latency. Obviously the alternative is to generate audio "just in time"
with a tiny chunk size, but this may have a negative impact on
performance.

Note also that direct mapping is not a feasible choice on some
hardware/scenario.

However this is off topic.

> important. If OSS had looked more like the IRIX audio I/O library, I
> think ALSA would look very different today, for example. That library
> only supports float32 data, for example.

Though the IRIX adoption does not imply it's the right choice...

-- 
Abramo Bagnara                       mailto:abramo alsa-project org

Opera Unica                          Phone: +39.546.656023
Via Emilia Interna, 140
48014 Castel Bolognese (RA) - Italy

ALSA project               http://www.alsa-project.org
It sounds good!