Re: OVRDBF in ILE (was: diff btw Procedures and Routines)

On Thu, 22 Aug 2002, Ken Sims wrote:
>
> Reasonable only if you use programs for menus.  We don't, we use display
> file type *MENUs.  However only in a few special cases are there *NEW
> programs with another *NEW program downstream, and that is where the called
> program is not called repeatedly, it is a program that is both a menu
> option and called by a function key from another program.
>

I'm sorry, but it's completely reasonable to have the first program called
by a menu be *NEW, and anything that it calls be *CALLER.

Since you seem very sure that *NEW is hurting performance, I did some
benchmarks.   Please keep in mind that I work for a small company, and our
AS/400s are at the low-end of the performance spectrum:

1000 iterations of calling a *NEW RPG IV program:

Machine:         Arch  OSRel Seconds Used    Time Per Call
---------------  ----  ----- ------------    -------------
9402-200(2031)   CISC   V3R2         159           0.159
9406-270(2248)   RISC   V4R5          12           0.012


The program that was called did nothing but set on *INLR and end.  The
times shown are the total times for 1000 calls to that program in a
simple DO loop.

This means that your interactive users, even if they're using an old
slow CISC machine, are wasting a total time of 1/6 of a second.  ONE SIXTH
OF A SECOND calling a *NEW program that could've been saved calling a
named activation group.

On the modern system, that figure was more like 1/100th of a second.
And that's on a machine that's near the bottom of the speed spectrum for
modern systems.

I'd be more worried about the time your employees are spending doing
things like blinking their eyes or breathing than I would about the
performance of *NEW activation groups.