Re: RPG/Java interaction

Hi Mark,

That's why I said the following in my first reply of the thread:

> If the Java program isn't throwing an error, it doesn't matter what
> you do in the RPG program -- the problem is on the Java side. But if > it is throwing the error, and RUNJVA is ignoring it, then you could,
> indeed, write an RPG program to solve it.

In your case, the program would definitely throw an error, so an RPG wrapper would work.

I can't tell you why RUNJVA works the way it does.

Indeed, I phrased things that way because I'm not overly familiar with RUNJVA -- I am familiar with running Java from RPG, so I pointed out that if that is the problem, you could write an RPG wrapper. In your case, that would work.

If you don't like how RUNJVA works, or want to know it's rationale, I'm not the person to ask. Maybe ask IBM.



On 2/23/2022 6:59 PM, Mark Waterbury wrote:

Scott,
I ran this command on IBM i 7.3:
    SBMJOB CMD('RUNJVA CLASS(garf.jar) PARM(barf)') +        JOBD(QGPL/QBATCHLOG) JOBQ(QGPL/QBATCH)  +        JOB(JAVABOMB)
where that class "garf.jar" does not exist ...  (on purpose).
The job ended "normally" but if I look at the job log, I see this:
Message . . . . :  -RUNJVA CLASS(garf.jar) PARM(barf)                             JVA0122    Escape                  00   02/22/22  13:33:36.354771  QJVAUTLJVM   QSYS        *STMT    QCMD        QSYS        01C8From module . . . . . . . . :   QJVAUTLJVM
From procedure  . . . . . . :   JvaSendMsg                                                  Statement . . . . . . . . . :   10                                                          Message . . . . :   Java program completed with exit code 1
Yet the job ended "normally."
Seems something is very wrong here?
Perhaps a simple CL "wrapper" that can do a MONMSG for JVA0122 and then throw a CPF9898 *ESCAPE message, to cause the job to end abnormally?
But why should that be necessary?