On Thu, Feb 04, 2010 at 02:42:33PM +0000, Jacob Meuser wrote: > On Thu, Feb 04, 2010 at 10:20:40AM +0100, Alexandre Ratchov wrote: > > On Wed, Feb 03, 2010 at 04:10:48PM +0000, Jacob Meuser wrote: > > > I wonder if these problems would go away if jack support were dropped > > > from portaudio. only working with 32-bit sounds suspicious. > > > > > > > that's easy to check by disabling it, I'll try it.. > > > > AFAIU, ALSA supports any format and seldom triggers > > conversion code of audacity and/or portaudio. > > > > sndio backend uses sio_getacap() to return native parameters > > and lets portaudio/audacity do the necessary conversions, > > since it claims it can. As far as I understand portaudio > > internals, this is the ``right thing''. So I bet we trigger > > various bugs. > > > > I don't exclude there's a bug in sndio backend though. But, > > I wonder whether we shouldn't just claim that any format is > > supported and stop using sio_getcap() here. > > that might cause other problems if aucat isn't used?
that shouldn't be a problem. Portaudio exposes the result of sio_getpar() to allow the app to display a nice drop-down list with the supported parameters. But if sio_setpar() fails, it will setup the appropriate conversion layer. AFAIU, audacity is misusing this mechanism. Not sure yet. I'm trying this right now... > maybe there > should be another arg for sio_getcap(), to choose between "native" > and "usable" paramters? > yeah, if we keep using sio_getcap() this seems necessary. -- Alexandre