Excel/POI - Please Help The Drowning Man

Arggghhh! I'm so close, but...

My RPG program reads in 1 or more .xls files from the IFS. The original 
problem had to do with the java environment exception not getting 
'cleared' if a given expected .xls file wasn't there. Scott Klement's 
workaround suggestion was to use the "access()" API to check for existence 
before attempting the read. That worked great. (Thank you, Scott.) 

He also said the real 'fix' would be to use ExceptionClear() to reset 
whatever is causing subsequent reads to fail. Well, now I need it. I got 
an "HSSF_open" failure this morning while trying to read the first 
EXISTING .xls file in the path. No idea why. When I immediately ran it 
again using debug, the first file read successfully. No idea why. Then the 
read of the second file failed. (Again, N.I.W.)

Once any kind of exception occurs in the environment, subsequent "HSSF" 
procs fail. Even if I run a different POI-related program, the 
"HSSF-begin" fails. I'm thinking that the only way these error conditions 
get cleared is when all our subsystems come down and restart at night.

I tried the following: ("envPtr" is defined as a pointer)

envPtr = hssf_get_jni_env()     (This worked after i added its EXPORT 
entry to the QSRVSRC member and re-compiled the service program)
ExceptionClear(envPtr)          (I can't get this to work at all. It says 
"pointer not set" or somesuch)

Can anyone help??????? I'm so close!

Arthur J. Marino
Southern Container Corporation
(631) 231-0400 x133