Hi,
a little background:
After some debate (on the linuxppc-dev list) I have been considering making
a "pedantic" version of the linuxppc pmac dmasound driver - one that _only_
provides the minimum that OSS requires.... actually, much more akin to the
ALSA way - we are on our way there... ;)
At the moment the (kernel, OSS) driver is somewhat bloated with code that
supplies soft emulation of rates/sizes and so on, that are not native to the
pmac sound hardware.
>From my reading of OSS programmer's guide (1.1) - it seems that sound apps
*should* *not* make any assumptions about what the soundcard can provide.
Of course, many apps do (make assumptions)... and thus the emulation of
these things - which eases support for apps without 'fixing' them.
Well, we are thinking that it is time to fix the apps - and make the kernel
stuff leaner (and easier to maintain).
----
so my question:
It is not clear to me from the guide whether:
(a) I *must* supply /dev/audio (at whatever rates/sizes the hardware
supports - i.e. a clone of /dev/dsp) ?
(b) If the hardware cannot do 8 bit, ulaw, 8K - I should leave /dev/audio
out?
(c) I'm obliged to emulate 8 bit, ulaw, 8K whether I like it or not?
tia,
Iain.
