On my Linux system, full duplex audio is achieved
by using two device files; one for input, and one
for output.  

As Csound comes out of the box, it selects /dev/dsp0
for both input and output.  This works, but not at
the same time.  In the past, I have altered the source
code to point to /dev/dsp1 for audio input, and all
has worked as expected.  I can use a microphone for
input, and hear csound-altered output at the same time.

I would really like it if there were a (dare I say it?)
flag or other method whereby one could configure the 
input and output devices.  There are reasons for this
other than just full duplex.  Perhaps I have two sound
cards working, and one plays cleanly while the other one
records better.

BTW, why can't I pipe csound's output samples into another
process?  It claims 'stdout audio not supported' or something.
There are some nice purposes for this as well.

Toby