Bear in mind, one of the reasons that IBM developed ILE in the first place was
to improve runtime performance of "highly modular" designs. If you were to
write an application that made use of modular design, but implemented as
program objects, you would eventually find that the dynamic call model was
degrading your performance. This is hardly noticeable if your application make
only a few calls per transaction, but if you were making hundreds of calls,
performance suffers. Think Synon/2E, and you'll have the idea. Synon wrote a
program for each atomic function in the action diagram. Some applications
would generate hundreds (or thousands) of programs. The applications were
modular, and functionally robust, but were painfully slow at times.
There were workarounds for this, like returning without LR, so that the program
was left in memory, but then you're left with the cleanup of resources at the
end of the run.... ILE brings relief from all this.
Static binding releives much of the overhead associated with a program call,
and activation groups help to manage the cleanup of resources. In addition,
since ILE compilers exist for most languages on System i, you can make use of
other language features that cannot be easily accomplished in RPG. The C
runtime is a good example....
As always, the real question is "what are the goals of the organization". IBM
has, for years now, published the "Developer's roadmap" which details a means
of adopting new technology into legacy applications. ILE is step one on the
roadmap. As described by IBM, this phase is all about modularity and code
reuse. Most of the remainder of the roadmap involves the use of these ILE
modules within new architectures, such as SOA (web services).
jmo,
Eric
-----Original Message-----
From: rpg400-l-bounces@xxxxxxxxxxxx
[
mailto:rpg400-l-bounces@xxxxxxxxxxxx]On Behalf Of
DennisRootes@xxxxxxxxxxx
Sent: Monday, April 23, 2007 4:09 PM
To: rpg400-l@xxxxxxxxxxxx
Subject: Help! Boss wants to know why ILE?
Please help me come up with some good examples as to what makes ILE better
than "business as usual". Here's what I've used so far:
1. It's faster - Rebuttal: doesn't matter we have plenty of cpu and our
machine screams as it is.
2. Service program equals reusability - Rebuttal: we can just use a
separate program for reusable code.
3. Local variables - Rebuttal: if they are inside a separate program it
doesn't matter.
4. System maintenance - Rebuttal: instead of service programs or
subprocedures we have separate pgms so it's the same thing.
It's not that the boss wants to stop us from using ILE, she just wants to
know what makes it so much better than plain 'ol RPG IV that she should
invest the man hours it's going to take to bring our whole dept up to
speed on ILE. And I just am not coming up with anything that's very
convincing.
So please, if you have any good arguments, let me know.
Thanks,
Dennis
midrange.com
