RE: CRTDUPOBJ an ILE program

[Scott Klement <klemscot@xxxxxxxxxxxx>]

On Tue, 11 Mar 2003, Buck Calabro wrote:
>
> Your benchmarks use very little storage.  That is, the called program
> doesn't have to initialise anything.  In a more typical RPG environment,
> there are files to be opened, variables to initialise and so forth.
>

But, it would have to do that part (opening files, initializing, etc)
even if you were running in the default AG, or even if you used QILE.

Unless, of course, you ended it last time with *INLR = *OFF.   In which
case, I certainly would certainly agree that *NEW is a bad idea, and you
should instead be using *CALLER.

> My original comment was based on practical experience with large-ish RPG
> programs in the CISC days.

My benchmarks included a (very low-end) CISC machine as well.   I don't
think the size of the program is relevant, because (again) it would have
to load the program no matter what activation group you used.

The benchmark is intended to show the time it takes to create / tear down
an activation group -- not to benchmark program startup times -- since you
have to start the program no matter what the environment is.

> Looking at your numbers (again!) makes me realise I should probably
> build some test jigs and get more modern numbers before I muddle the
> issue more.  I have no *NEW programs that I can quickly run tests on.

I agree.