At Mon, 19 Jun 2006 22:12:57 +0100,
Alan Horstmann wrote:
>
> > > I have devised an
> > > alternative using the existing SNDCTL_DSP_CHANNELS, by passing 'channels'
> > > calculated as (playback_channels + 256 * capture_channels) so that in
> > > effect the first 8 bits are playback, the 2nd 8 bits are capture. The
> > > result is compatible with the existing, as numbers < 255 are treated in
> > > the existing way, setting playback and capture the same if the
> > > sub-streams are active. Thus it has simply added an enhanced mode.
> > >
> > > Diff is attached; I wonder what your thoughts are on this?? I have been
> > > able to very simply patch portaudio to use this facility and the system
> > > now works great.
> >
> > I don't think it's a right solution. This simply introduces an
> > incompatibility with the existing OSS ioctl...
>
> The 'enhanced' alsa-emulation I created is backwards compatible; it enables
> aware apps to work-around the limitation of the emulation. But if sent
> commands as-if standard, does exactly what it would have done. However,
> where the app needs to set separate record and playback channels for a duplex
> mode, there is extra capability. It would be used only for duplex when
> (capture_channels != playback_channels) This would not be necessary with
> original OSS, although also it would not work. None-the-less I can
> understand your reluctance, it is just that a solution seems necessary.
>
> Another alternative would be for ALSA to create at least two devices, one for
> all playback channels, the other for all capture, eg /dsp0 and /adsp0 or
> something else. It would then be more like the raws OSS in its behaviour.
> That would also solve the problems here. Obviously a full implementation as
> per attached description would also be fine!
It's easy over aoss. Having the following
pcm.dsp0 {
type asym
playback.pcm "plughw"
}
pcm.dsp1 {
type asym
capture.pcm "plughw"
}
would give you /dev/dsp0 for playback and /dev/dsp1 for capture.
You can replace "plughw" whatever you like, of course.
Takashi
_______________________________________________
Alsa-devel mailing list
Alsa-devel@xxxxxxxxxxxxxxxxxxxxx
https://lists.sourceforge.net/lists/listinfo/alsa-devel
