On Sat, Jun 18, 2005 at 02:34:46AM +0200, Christoph Eckert wrote:
I think we should (and can) keep the desktop and 'pro'
worlds separate.
I do not agree :) . We're in the free software world, so
there's no need to tell the non-pro-audio-users "use anything
else".
How OS X solves this problem may be instructive.
[most facts below are probably right, I'm sure I got
a few wrong ...]
In the System Preferences->Sound menu, a user
can choose the default audio input and audio output
device.
In the CoreAudio API, this choice becomes the
current value of kAudioHardwarePropertyDefaultInputDevice
and kAudioHardwarePropertyDefaultOutputDevice.
Consumer-oriented apps use those Defaults, as
a rule; content-creation apps usually have their
own Preferences that let you select the audio
devices for the app.
CoreMIDI works this way too.
So, the underlying system is "pro", but there
are provisions in the API to handle the mindset
of both pro and consumer worlds.
Finally, note that Apple solved the legacy problem
by emulating Sound Manager (OS 9 audio API).
Until Tiger, QuickTime actually used Sound Manager.
Many other apps still do.
If Linux followed this model, one would write emulators
over jack for all of the consumer audio APIs being used in
jack, to accommodate the installed base, and have
the emulations work well enough to convince distros
to just ship jack as the bottom layer, and use its
emulated APIs. Then, you can start evangelizing
direct jack calls to all application developers, consumer
and pro alike.
Yes, this takes a lot of work, and is not in the direct
path of solving pro audio problems. Just like I've spent
5 years on pushing RTP MIDI through the standards
process because I felt someone had to do it, I think the
Linux audio API problem only gets solved if someone
decides to dedicate 5 years of their life to doing it.
That's how long a systems project takes to have
an impact on the world. Good luck.
---
John Lazzaro
http://www.cs.berkeley.edu/~lazzaro
lazzaro [at] cs [dot] berkeley [dot] edu
---
