RE: Beyond RPG CGI

> From: Joel R. Cochran

I just want to provide an alternate viewpoint to some of Joel's opinions.
This is not to say that any of his views are invalid, but that I've got some
slightly different experience in some of these areas.

> 1.  Understand that Java is way more than just another
> programming language,
> it is really a platform (or a sub-platform).

Yes and no.  The language, Java, is extremely powerful in and of itself.
All the other bits and pieces, especially those in the J2EE standard, are
powerful but highly, highly optional.  Depending on the work you do, you may
never need to code an Enterprise Java Bean, yet you may make great use of
Java.


> 2.  If you are going to learn Java, then I suggest you learn
> Java... what I
> mean by this is that many people start off trying to use an IDE such as
> VisualAge for Java, JBorland, or Forte Community Edition.

Yes, learn to use a command line to build some basic classes.  This should
take you a few weeks.  Then immediately switch to Visual Age for Java for as
long as it's around.  VAJ is simply the best Java IDE available.  I don't
use the Visual Composition or any of that other crap, I just like the IDE
because it's built for Java developers - you can easily move methods around
from class to class, you can view the entire source, impact of changes are
immediately visible, and the ability to see who references what is
outstanding.

Unfortunately, VAJ is dying.  I'll be very interested to see if the new
Eclipse-based editor can hold up its end of the stick.  The other IDEs are
basically just revamped C and C++ editors and really have no business in a
Java development environment.


> 4.  Personally (and I'm putting an umbrella over my head to protect myself
> from the coming hailstorm) I would avoid WebSphere.

If you plan on doing anything on the Web (that is, servlets and JavaServer
Pages) then you are doing yourself a disservice to not take advantage of
WebSphere.  It's much like any IBM product - incredibly flexible, but a bear
to master.  However, you don't NEED to master it, just learn a few basics.
If you can afford having someone come in for a couple of days to confgure
your system and show you what you need to know, then by all means do it.

Tomcat and all the other free goodies are fine, but in the long run you may
find yourself spending a lot of time trying to keep up with the current
version and making sure that the various pieces work together.  WebSphere is
free, and just as powerful as Tomcat.


> 6.  You are way ahead of the game learning JSP/Servlets if you
> already know CGI.

Yes and no.  It depends on what you want to do.  CGI is nothing more than
outputting HTML with a program.  You need to fully understand how HTML works
and CGI can help.  However, if you understand the basic syntax of Java, you
can do a ton of stuff with servlets and JSPs and never worry about CGI.


> 7.  Finally, from one RPG programmer to another... learn JDBC early.

This is where I disagree the most.  JDBC is an interpreted SQL.  It's slow
and cumbersome, and has all the drawbacks of SQL without any of the
benefits, since you need to learn another language (Java) just to use it.

Java is not and probably never will be a business language.  EJBs are
overkill, and JDBC is a kludge.  Use servlets and JSPs to create a browser
interface, or Swing to create a thick client, and then communicate via the
messaging technique of your choice to an RPG back end.

Make no mistake: RPG is by far the better language for business logic.  Java
does not even come close.  I have worked with and taught Java for years and
sell two Java-based products, and I will still tell you that you shouldn't
write a single line of business logic in Java.

The reason is that OOP is not really a good technique to use for business
development.  Object oriented programming is great for encapsulating logic
that rarely changes, like formatting HTML, or processing XML.  But when it
comes to the more flexible requirements of business programming, OOP falls
down.  That's because no two companies can agree on business logic - heck,
they can hardly agree what a calendar does, much less what a company is.

Anyway, those are some thoughts from the other side of the fence.  Use 'em
or not.

Joe