ALSA [Was: Paris summit]

[Lee Revell]

On Wed, 2006-06-21 at 19:42 +0200, Jeff Waugh wrote:
> <quote who="Lee Revell">
> 
> > Um, sorry, I disagree.  ALSA supports the same read()/write() programming
> > model that OSS does (and several more).  What's your problem with it?
> 
> The overwhelming response I get from ISVs is that the libalsa interface is
> complicated - both for doing simple things and hard things. They worry about
> API/ABI stability. It's relatively unproven, in particular for ISVs... and
> that's what all the complaints are about -- Skype, Flash, TeamSpeak, etc. On
> the other hand, OSS is dead simple for an ISV, with zero dependencies. I do
> not blame ISVs for their choice, not for a minute. I have every sympathy.
> 

I can see where they are coming from.  I would say it's proven by the
fact that there's a large Linux pro audio community who all use ALSA +
JACK.

But I also think OSS is a very rude choice because it blocks all other
apps from using the soundcard.  They are massively inconveniencing users
because they can't be bothered to grok ALSA.

> > And what alternative do you suggest?
> 
> I've been working through the upstream desktop community and OSDL DTL to
> come to a useful conclusion for this. It is very hard, because there are a
> number of semi-conflicting use cases and requirements. However, I do *not*
> believe that libalsa is part of the answer, at least in terms of exposed
> interface (to ISVs and everyone else). I think a combination of JACK, Pulse,
> the GStreamer multimedia framework and a very simple audio API (something
> like libao) will eventually solve this. The challenge is getting it right
> and establishing buy-in across the community *and* ISVs... But that's the
> fun bit for me. :-)

Agreed, a higher level API would be good.  I am glad someone is working
with ISVs on it.

Lee