RE: Java and cleanup from exception

["Dennis Lovelady" <iseries@xxxxxxxxxxxx>]

Hi, Scott:

> > For the same years, I have found that when Java errors occur when
> > testing those functions ...the interactive job is virtually useless
> > to Java.  Cannot start a new Java session, and cannot recover the old
> > one.
>
> Hmmm.... that's not entirely true.  If you get an error from Java in
> RPG, you can typically run more Java without a problem.
>
> However, if you do something like pass a *NULL object reference to a
> Java function, you will cause the whole JVM to crash -- and _then_
> you'll be in the state that you describe, where Java is useless until
> you sign off or RRTJOB.

Yes, that was it (the *NULL object reference), and I had not picked up on
the fact that this (monitorable) condition is actually at the source of my
woes.  Ultimately, if a java function fails - resulting in *NULL return, my
code tends to go on to the next thing since there's no trap capability
between Java and RPG.  So the next statement (that uses the value that
resulted from the failure) is what crashes the JVM!  Interesting!  This, I
can deal with!  I'll make my own "trap" at points where it makes sense.

> After running RRTJOB  you can restart the JVM.  This is mildly less
> bothersome than signing off and on again.  (However, this RRTJOB trick
> doesn't seem to be documented anywhere -- so I wouldn't use it in a
> program.)

Nice trick.  Thanks!

> But the best thing to do is write your RPG so that when you get an
> error
> from Java, the program immediately aborts, and doesn't try to use the
> bad object reference that might've just been created.  I realize that
> it's easy to have bugs that don't follow this recommendation -- but if
> you make an effort to catch all of the *NULL object references before
> they get used, the "must restart the JVM" situation doesn't come up as
> often.
       
Yes, I see that now!  Since most of the stuff I write either has a wrapper
or could have a wrapper, this will be pretty easy (everything is relative)
to overcome.  Thanks again, Scott.

Dennis Lovelady
http://www.linkedin.com/in/dennislovelady
--
"An optimist proclaims that we live in the best of all possible worlds - the
pessimist fears this is true."
        -- James Branch Cabell