Last month, inspired by the 'floats' discussion on the list, I generated
some float files from csound and then ran those float files through
another csound instrument via the '-i' flag to scale them and write them
to an AIFF file. I was doing this work on Linux Csound version 3.482.1a.

I tried again this month using version 3.485, and it appears to me that
this newer version is not writing or reading float files correctly. When
reading the float files via the "-i" flag, csound is claiming that there
is no soundfile header and that it is assuming that data is shorts. It
then procedes to create a sound file of 100% tasty noise. 

I get the same results running Irix csound 3.485.

Here are the test scenarios:

----------
Scenario 1: 
reading an old float file I generated and used back in August

v3.482.1a: "reading 8192-byte blks of floats"
v3.485: "no soundfile header,assuming shorts"

-----------
Scenario 2:
reading in a fresh, new float file made with v 3.482.1a

v3.482.1a: "reading 8192-byte blks of floats"
v3.485: "no soundfile header,assuming shorts" 

----------
Scenario 3:
reading a new float file made with v3.485

v3.482.1a: "no soundfile header, assuming shorts"
v.3.485: "no soundfile header,assuming shorts"  
 

----------

Therefore: it seems csound 3.485 has a problem in BOTH reading and writing
float files. 

Any one else seen this? Am I missing something?

Thanks,

Bret Battey