Re: Divide by zero and monitor group...

[Scott Klement <rpg400-l@xxxxxxxxxxxxxxxx>]

Robert,

Monitor groups always log their errors to the job log (no matter what 
the error happens to be -- this is not unique to divide by zero.)

This can be very handy for troubleshooting problems... and is a feature 
I really appreciate.

I have a subprocedure in a service program that will remove the error 
messages from the job log.  So I can have some logic in there that if 
it's an error I'm expecting, I'll remove the messages (gets rid of 
useless clutter in the job log) but if it's an unexpected error, I'll 
leave it in there for diagnostic purposes.  Works very nicely.

You're right that this does have an impact on performance.  You should 
not use this to catch divide by zero errors if you expect them to happen 
frequently...   I usually use an 80/20 rule.  If I expect an error to 
occur less than 20% of the time, it's a candidate for a MONITOR.  If the 
error will occur more often than that, I write code that anticipates the 
error and fixes it in advance...

if myfield = 0;
   result = 0;
else;
   result = somefield / myfield;
endif;

If myField=0 is expected to happen freuently, this will run MUCH MUCH 
faster than monitoring for the condition...

-SK


On 7/1/2015 2:11 PM, Robert Rogerson wrote:
> Hi all,
>
> I was somewhat surprised to find that a divide by zero error inside a
> monitor group still wrote/added an error/message to the joblog. The
> message I'm referring to is
> MCH1211 Escape 40 Message . . . . :   Attempt made to divide by zero for
> fixed point operation.
>
> I was pointed to this error as the reason for the program abending, but
> when I checked (the specific line specified in the message) in the
> program it was within a monitor group, so it was already being handled.
> In the end the issue was elsewhere in the program.
>
> With that being said, my question is, doesn't writing to the joblog take
> up processing time and isn't this message misleading when the program is
> already handling the error within a monitor group?
> I know that checking the divisor before doing the divide operation would
> avoid this issue, but I thought this was one of the purposes of the
> monitor group.
>
> Thanks,
>
> Rob