RE: Divide by zero monitoring

["Wills, Mike N. (TC)" <MNWills@xxxxxxxxxxxxxx>]

They are in arrays... however since there is no easy way to print the array
contents to an externally described printer file, I am forced to have one
line of logic for each field to print. I would not have bothered asking the
question if I had it in a loop.

Unless you are talking about something else....


-----Original Message-----
From: Reeve Fritchman [mailto:reeve@ltl400.com]

Bob has a point.  Mike, if you can put everything into a couple of arrays,
you can eliminate the redundant coding of checking for a zero divisor.  Loop
through all the elements and check element by element...

-----Original Message-----
Behalf Of Bob Cozzi (RPGIV)

Let's not forget, that there is overhead involved in calling a
procedure!
I ran a test using the XLATE operation in my TOUPPER()
procedure/function. It took 26 times longer to run the code in a
procedure than it did as in-line code. Now, that's a couple lines of
code vs a procedure wrapper for that same few lines of code.  Of course
if you put 100 lines of code into a procedure, the overhead isn't going
to increase, it'll stay the same.

Until IBM adds a "Inline" keyword to procedures, use of procedures for
certain time-critical things (like math) might be better left on the
wish list.


Specs:

Called the procedure 30000 times
vs.
Inline XLATE and SUBST opcodes.

Procedure was approximately 14.17 units of measure.
Inline code was 0.02 units of measure.

So the overhead for a procedure call is X where X = approx. 13.8 units
of measure regardless of the size of the procedure.