Re: [fluid-dev] FluidSynth on Raspberry Pi with Wolfson Pi card

[R.L. Horn]

On Thu, 21 Apr 2016, Element Green wrote:

> Sometimes there are limitations with audio drivers in regards to sample
> rate, buffer sizes, etc.  You may want to try 48KHz or 44.1KHz (depending
> on which one you are using at the moment) for the sample rate.

I think that would result in a different error message (actually a warning 
and the synth playing out of tune), though it's still worth a try.  Buffer 
problems throw up a warning and result in terrible latency, but that 
should be all.

As it happens, "The audio device error: ..." only occurs in one function: 
fluid_alsa_handle_write_error() (thank goodness for bad English).  This 
function is only called after snd_pcm_writen() and snd_pcm_writei() calls. 
For some reason, those are returning EIO after apparently successful 
initialization.

I'd check the permissions in /dev/snd, but I don't think you'd get that 
far if those were a problem.

I'd also try a new alsa-lib, if one is available.  My inclination would be 
to build from alsa-project.org sources if possible.  Again, I don't think 
that will fix it, but what the heck?

Finally, try using plughw: instead of hw:.

> What other applications have you tried that have been successful?  For 
> example, aplay would be a simple way to test ALSA.

I'm wondering the same thing.  In particular, if aplay works, it would 
lead me to suspect that the driver is somehow failing to cope with 
non-blocking I/O (i.e. write() sets errno to EIO instead of EAGAIN).  But 
that's just a wild stab in the dark.