Java - the new COBOL (Revised)ΒΆ

date:2005-11-03 14:55:33
category:Economics of Software

Kontrawize says [link ] that Java is the new assembler.

/dev/null [link ] says “While we are willing to expose and discuss the problems in Java, we do it because we want to improve it, not because we need to cut & run”. Taken out of context, I could twist this to mean that Java is no longer looked at merely technology to build a solution, but has developed an independent existence and a rabidly patriotic following.

I think /dev/null is being a Java realist (“it has problems, but it solves some problems well”). I think Kontrawize’s point isn’t quite right; really the point is that Java is sometimes applied to the wrong problems.

While Java is the next big legacy, some of Kontrawize reasons are wrong. Comparing Java to Assembler as low-productivity toolsets, misses the mark. Assembler is only low-productivity if you’re using it for the wrong thing. If you’re writing I/O drivers, assembler rules. If you’re writing a multi-media presentation, assembler is the pits. It’s the “mismatch between levels of abstraction” problem.

In the IT shops I’ve been in over the last few decades, assembler is a legacy that leads to absolute deer-in-the-headlights paralysis. The few shops that still had some assembler had to treat it as a black box. The software had to be documented without reference to internal structure, and the replacement had to involve the potential for business change, since the nuances of the assembler were already lost forever.

COBOL, on the other hand, is more like Java. IT shops have lots of legacy COBOL folks, and when you try to introduce something new (like Java or Python) to them, they fight back every step of the way. They often cling to old work processes, unwilling to consider that the brave new world is upon them.

One example: We build the entire Java application system and turn over four JAR files that are put into production. The COBOL folks ask questions like:

  • Why turn over everything? Why not just turn over the “program” which changed?
  • Why recompile everything? Wouldn’t be “simpler” (“more efficient” was also used) to recompile only the “program” that changed?
  • Why don’t you have proper application main programs? Why do you have these reusable packages that are recombined every which way? Wouldn’t it be “simpler” to use “copy books” (a/k/a/ include files) so you could more easily track the dependencies among your programs and the reusable bits?
  • What’s wrong with shell scripts? They’re just like JCL and we use JCL all the time.

I think the idea that Java will become the next COBOL is very important. One consequence will be the “Technology Freeze Play”: we’re an All-Singing-All-Dancing-All-Java shop and you can take your silly Python (or Open Source components) and just stick it in your ear. We measure effectiveness by our entrenched skill set, not cost to solve a business problem and value of the business problem being solved.

When I taught C programming, I remember one conversation about upgrading the skills of some of the COBOL people in addition to bringing in new C programmers. “We don’t really plan on upgrading our COBOL programmer’s skills”, I was told. My response: “Why not? Were they born stupid?”

Addendum.

Is it arrogant to suggest that COBOL programmers can or perhaps should learn C?

Or, is it arrogant to suggest that COBOL programmer should not be pigeon-holed?

Or, is it arrogant to suggest that pigeon-holing COBOL programers is the same as labeling them “born stupid” and, therefore, untrainable?

It is probably arrogant to suggest that COBOL programmers are being maliciously held back from learning new technology. Sometimes, they aren’t offered the training because of the huge legacy maintenance burden that they have to support while new technology is being introduced elsewhere.

This, I think, is why Java will become like COBOL – a legacy which we hate but must support. The people supporting Java will be pigeon-holed. New technology efforts will be focused elsewhere, leaving these people behind. Not because they’re literally born stupid, but because they’ve had the bad luck to be supporting the technology that became the reviled legacy.

Previous topic

Paying Micro$oft Twice

Next topic

More Marketplaces

This Page