RE: How to end RPG program abnormally?

>Buck, the question isn't clear to me.  What is 
>the problem with the Halt Indicators?  

They'll cause a white message to be thrown, stopping the job.  These are
essentially batch programs that can never ever throw a message that will
stop the job.  The deal here is that a Java program will be running in a web
server and it will call a handful of RPG programs to go fetch data as well
as fire off some other work (a la SBMJOB).

>I'd hate to recommend using U1 thru U8 but 
>possibly that'd help???  

Not really.  I don't think Java can inspect those indicators, but it's an
interesting question to research.

>What has to happen to make a Java call fail?

Basically the same thing that would turn %ERROR on in an RPG program.  The
idea is that Java does a CALL to my RPG code, which does its thing and
returns (never throwing a message that would stop the job!)  What our Java
programmer wants is for all server-side errors to act exactly like the
program DID throw a message and die.  Same behaviour you'd get if you called
an RPG from a CL and the RPG did a divide by zero.  Once you cancel the RPG
error, the CL gets an error to the effect that the call to the RPG program
failed.  I can MONMSG that.

Our Java programmer is doing the same thing.  Essentially she has a 
CALL RPG
MONMSG 

And she either wants the RPG-returned data to be perfect, or the MONMSG to
fire.
  --buck