OO COBOL: Overall Impressions

IBM's version of OO COBOL is both good news and bad news.

The good news is that it's so much simpler than C++. Because there are fewer options, there are fewer choices to make, and fewer ways in which those choices can interact. The bizarre subtleties that plague C++ programmers will be much less common in COBOL.

The bad news is that it's so much simpler than C++. The machinery of encapsulation and inheritance are too primitive to express more than the most rudimentary classes.

Some C++ features you can do without. For example, you don't need public data members -- you can always define public methods to access private members.

Other missing features you can't fake. In particular:

You can't define static members;

You can't define private or protected member functions;

You can't declare one class as a friend of another.

You also can't declare protected data, but you wouldn't need to if you could define protected methods.

It is often useful to circumvent the encapsulation of a class. C++ lets you do so in a carefully controlled manner, through the use of friend classes, friend functions, and protected members.

By contrast, OO COBOL's encapsulation is so rigid as to be self-defeating. The only way to provide any access to private data is to define public methods, which render the data effectively public.

Still, it would be churlish to write off IBM's OO COBOL as a failed experiment. The draft standard promises a fuller set of features, and IBM may catch up eventually. Meanwhile, even the limited features now available are worth exploring, and will no doubt prove useful.

Cobol Home

OO Cobol