I am continuing to look at LibSndFile and the opcodes diskin and in.
Diskin seems OK in my rather scrappy tests, but I am continuing to
have strange problems with in.  It occurred to me in this that there
may be a conflict between what LibSndFile would like and PortAudio.
In particular LinSndFile counts in frames, not samples and certainly
not bytes as Csound4 does.  My current code is moving increasingly to
frames, and has abandoned bytes.   So audrecv fills a MYFLT buffer
with n samples.  This may be in conflict with what the values rtrecord
expects. As I am now totally out of sync with Michael's code I do not
know how to resolve this, or even discover if it is a problem.

This not directly related to the problems I am having with in, which
seems to be crashing in sf_read_float when reading the second
buffer-full and I cannot understand why.

==John ffitch

>>>>> "jpff" == jpff   writes:

 jpff> This not directly related to the problems I am having with in, which
 jpff> seems to be crashing in sf_read_float when reading the second
 jpff> buffer-full and I cannot understand why.

Now fixed.  Hurrah

==John ffitch