Re: Re writing CL in RPG

Let's not assume that IBM can write their command programs any faster than
your average RPG programmer. Using Scott's IFS tutorial has shown me that
we can often blow the socks off of IBM. His IFS stuff is much faster than
CPYTOIMPF or CPYFRMIMPF. Then again, if you ran either of those commands
underneath debug you would see some intense SQL going on.

Then again, there's not just speed, but granularity to consider. What
happens, if on record 2,125,432 if there's a constraint violation or some
such thing? With RPG your write fails, you write it to an exception
report instead, and maybe a file to be reposted. (Well, as long as you're
not like some clowns who assume that any write violation means that the
record exists and you try an UPDATE instead.)

Another thing to consider, if the source files can be read in key order,
you may not get the blocking on the output to the summary file, but the
summary file will then be in key order. Well, if you really want to
ensure output in key order, perhaps some matching records logic, either by
cycle or by hand. That might help the speed of using the summary file
afterwards. Then again, that could easily be fixed by RGZPFM keyfile(...)
to sort the file afterwards. But if it is one of those 24x7 files...

Rob Berendt