[LAU] Zyn and the art of jack audio/midi

[hollunder at gmx.at]

On Sat, 01 Aug 2009 12:52:44 +0200
Atte Andre Jensen <atte.jensen at gmail.com> wrote:

> cal wrote:
> 
> > There's an experimental patch against the 2.4.0 release tarball at
> > <http://www.graggrag.com/?q=node/19>.
> 
> I played a bit with it, and it seems to work fine. I only selected 
> patches and played them on the keyboard, though. Someone (Will
> Godfrey, perhaps) should stress test it with heavy load and see wht
> happens.
> 
> A thing that struck me as odd: Why do I have to specify my jack 
> buffersize, when zyn obviously knows what it should be:
> 
> [atte at vestbjerg ~]$ zynaddsubfx
> <snip>
> Sample Rate = 		48000
> Sound Buffer Size = 	256 samples
> Internal latency = 	5.3 ms
> ADsynth Oscil.Size = 	1024 samples
> 
> AudioOut, zynNframes 256, zynSamplerate 48000
> Error, zynNframes 256 != jack buffer size 128 => bad
> Error, failed to open audio out on 'default'
> 
> So instead of me having to go "zynaddsubfx -b 128", the program
> should know that the only sane buffersize with the running jack
> server is 128 and then use that.
> 
> I'm not sure if this behaviour is new, but I don't seem to remember
> this...
> 

I think there always was an external and an internal buffersize. But
yeah, it should be trivial to get jacks buffersize and use it.

Less trivial might be on-the-fly buffersize changes. Those are possible
since a long time but few applications support it currently, in
hydrogen for example it only works in very recent svn revisions.
This is not a deal breaker, sane and stable jack performance is far
more important.

Philipp