J2EE is not complex. Distributed Computing is Complex!

[nhebert]

All,

To Quote:

"Java Adoption Hampered by J2EE Complexity,
According to New METAspectrum Market Evaluation"

Bullshit!

J2EE (and EJB) is not in and of itself complex.

The general concept of DISTRIBUTED COMPUTING is complex!

In a previous life I spent 6 years as a Forte Software
Consultant. In my mind, Forte examplified and realised
some of the finest concepts in Distributed Computing.
I still have heaps of satisfied customers in Australia to
prove this.

As a Forte Consultant and ad hoc Trainer, I often found
that the Forte TOOL language and the Forte Runtime was
not so hard to understand as was the *concept* of
computing or processing in more than one address space.

Now as a J2EE Technology Consultant, it is no different.

I have discovered this simple fact:

Most people simply do not or can not think outside of
the single address space box.

Most J2EE professionals I have encountered have not
deployed an application across multiple application
servers. No, not AppX on node A, B and C. Thats a single
address space times n. Too easy.

I'm talking about one copy of AppX-part1 on node A,
one copy of Appx-part2 on node B and clustered copies
of AppX-part3 on nodes X, Y and Z in a fail-over mode.
Now *THAT* is an example of distributed computing.

Mapping the concept of a single logical application
across multiple physical nodes, simply does not
register with most people.

This is not a problem of J2EE.

Toss this one to a .NET "Architect". ;-)

Would like to hear what you all think about this.

Cheers,

Noel.

[schaefera]

Hi

My experience with J2EE is too complex is that developers and architects don't understand that J2EE is a framework and must be understood in order to use it most effectivly.

It is stupid to use Entity Beans when you want to aggregate functions on the data (sum, average etc.) but I think it is stupid as well not to use Entity Beans when you deal with records on a DB.

BTW don't forget that J2EE never talks about clustering. This is just a concept underneath J2EE.

Have fun - Andy

[dsundstrom]

> It is stupid to use Entity Beans when you want to
> aggregate functions on the data (sum, average etc.)

Unless you are using the EJB-QL 2.1 spec which has these functions. Of course we don't support 2.1 spec yet.

[xdariox]

Yes, Distributed programming is difficult, and EJB does a good job at hiding the complexity. In fact I think the problem with complexity in EJB is not the distributed aspect of it. There are two main reasons for the complexity in EJB:

1) I belive the complexity in the EJB specification stems from the fact that it mixes distributed computing with Object persistence. Had it just concentrated on distributed computing it would have been lot easier to deal with.

2) I also feel the other complexity in EJB, as is the trend with the current software, is its reliance on XML type declarations. Your application is in Java and XML. We all know how to debug a piece of code, but how do you debug a XML deployment descriptor. A bug in your application may very well be a bug in your deployment descriptor, yet there are no way to debug it.

Daryoush