RE: Locking records in DB2

> From: Paul Holm
> 
> Perhaps with your views, you should let the software industry know the
OO
> /
> frameworks don't really work for real and complex business logic
because
> the industry is investing heavily in these technologies.

Paul, I've been doing just that for several years now.  I think I've
made it clear that I believe procedural languages and particularly RPG
are much better for developing complex, data-driven business logic than
OO, especially strictly typed OO languages, and I've actually shown that
in a number of benchmarks.

Not only that, but history has shown that OO frameworks are difficult to
design -- the San Francisco project is just one massive example of how
difficult it is to translate the complex, evolving requirements of a
business to a rigid hierarchy.

This debate has in fact been raging in the industry for some time, and
is why there are specific camps even in the OO world.  The two biggest
competing ideologies today are the strict typing group, which includes
adherents of languages such as Java, and the soft typing group that
promotes languages like Python.


> I, for one, believe they do.  A properly built framework provides
common
> services and allows for exceptions and complex editing to be handled.

Ah, herein lies the rub.  The modifier "properly built" is huge, and
frankly there are few systems of more than moderate complexity that are
properly built to handle changing business needs.  The problem is in
object hierarchy definition.  While there is no argument as to what an
HTML tag looks like, there is no fundamental agreement on what makes up
a customer.  Heck, there are differences as to what makes up a calendar.
And even in the same company, what makes up a customer today may not be
what defines a customer tomorrow.  And because of this, there has been
no successful general business framework that comes even close to the
complexity of a modern ERP package.

It's not impossible, but it hasn't been done yet.  And the question is
continuously obfuscated by those who insist that data entry applications
represent business logic.  They are the very lowest form of business
logic, and barely scratch the surface of what a typical ERP package will
do.

That's why I have (for about three years now) held out the same
consistent challenge: write an MRP generation module in an OO language
that matches the performance of an RPG solution.

Once that has been done, then I think we'll see the first properly built
business framework.  Until then, it's just another round of DFU.

Joe