RE: FW: Java Vs RPG on iSeries

Blair, I'm not going to dispute much of what you say.  I'll just give my
viewpoint, because that's probably the biggest difference between us.

First, my background taints my view.  I've been a midrange business
application developer for over 20 years, and the problem space I've in is
primarily the sort of database intensive applications used that are used to
run real, live businesses.  My particular focus was always ERP systems, one
of the traditional domains of the iSeries and its predecessors, all the way
back to the venerable S/3.  So when I talk about business applications, and
particularly legacy business applications, that's the space I'm talking
about.

Java does NOT play well in that space yet.  And when traditional iSeries
shops talk about business applications, these are the applications they're
talking about.  So, no, Java does not do as well in what I consider to be
business applications.  I don't know of a single shop that needs "full-color
JPEG visualizations of trends
in the MRP data".  Honest I don't.

And I honestly don't worry much about SPECjbb2000.  I worry about MRP.  And
not only that, I worry about MRP as programmed by people who understand it
from a business standpoint.  You may be right that you could apply some
wonderful parallel processing techniques to MRP to speed it up.  But that's
supposition, and takes for granted the rather sophisticated programming
expertise required to develop truly parallel systems - expertise not
generally available in ANY market, much less the iSeries market.  Until you
can deliver a programming environment where a workaday programmer can do the
same job more efficiently in Java, it's all academic and pretty useless to
somebody running their shop on a model 270.  And that's the real world, or
at least my real world.

So I think we can agree that while SPECjbb2000 numbers are impressive, they
don't make much difference to the majority of iSeries customers.  Maybe I'm
wrong - if so, I'm sure you'll tell me so <smile>.

So here's my take on it:

While Java makes us think in parallel ways, what difference does that make
to someone posting their general ledger?  What difference does it make to
someone allocating shop floor inventory?  These "meat and potato"
applications are the ones that have pretty much built the midrange market,
and are the ones that seem to get shunted aside by the newer technologies.
In the mad rush to web-enabled graphical data mining applications for
multinational corporations, the 100-man manufacturing shop is getting short
shrift, because frankly none of these tools do him a bit of good.  Most of
them don't work and play with his legacy systems and databases - systems
that have been serving him and his company for five, ten, fifteen or more
years.

Technology is great, new applications are cool, but if you leave the old
customers behind, you're forgetting the people who paid for all this new
stuff in the first place.  If it wasn't for the S/3, S/34, S/36 and S/38
users, you wouldn't have an iSeries to play with.  And what Java can do in a
massively parallel 32-way 890 doesn't mean diddly to a guy who's trying to
keep 100 people employed.

I'm trying to get Java into THOSE shops.  So that they are positioned to
take advantage of the next wave of applications.  So that they can
web-enable their systems without having to redesign their database or hire
an entirely new development staff.  If the meat and potatoes shops embrace
Java, you'll see it grow a lot faster than if only the biggest corporations
do.

I guess I just am focused on the little guys and what we can do for them,
since they've basically paid my salary for a long time.  And what we can do
for them is to make a real, usable architecture that allows them to use Java
as a front end to their legacy applications.  Once they get that in-house,
they can use that experience as a bridge to the next generation of
applications that may be entirely written in Java and SQL.  But that
requires people who recognize the strengths and weaknesses of ALL the
various tools, and aren't afraid to acknowledge that RPG still does some
things better, and in fact the things RPG does do better may be more
important to a lot of shops than the latest data-mining algorithm.

Joe

P.S. As to me having to prove my assertion that RPG is faster than Java: I
sort of understand your point, but to me it's like owning the heavyweight
title.  My past benchmarks proved RPG faster than Java.  Until Java takes
the crown with a win, RPG retains the title.