When I used to ride the train to work, I sat next to a young man who was looking over a printout of what I recognized as COBOL code. I talked to him, and he was debugging a problem that came up the night before in the back-end of a gas pump point-of-sale system. He was very cheerful and seemed to love his work. After that I thought about that one nice thing about being a COBOL programmer was that you didn't have to keep up with all of the latest frameworks and standards.
Reading this article made me think of Jared Diamond's assertion that invention is the mother of necessity, not the other way around.<p>Before anyone gripes about outdated technology you need to ask yourself a question: what's so great about the new stuff? I'm not saying there isn't an answer to that question (I'm also not saying that there <i>is</i>), what I'm saying is that the answer needs to be better communicated before anything is going to change.
It seems that there a few broad problems with owning/using legacy COBOL:<p>1. It's usually run on a big, expensive mainframe.<p>2. Good developers run from it, so you're stuck with the coders left over from adverse selection<p>3. It has questionable modularity/maintainability and integration with more modern stuff is hard.<p>What opportunities are created by these problems?<p>My limited experience is that Mainframe COBOL systems (I know nothing of Microfocus) use either a database (DB2 usually) or large, random access flat files for persistence. COBOL programs are much like scripts run from within a transaction monitor. [Please correct me if I'm wrong here!]. Why has no one done a WINE-style implementation of the important APIs along with an interpreter so that existing code can be run on a horizontally-scalable platform? Mainframes are really expensive. Not only does IBM charge a lot, but you end up buying 3rd party software (ABEND-AID, I'm not kidding...) just to keep the lights on. Perhaps a distributed flat file implementation could be built to mimic 60's era data storage models. Does anyone do this? It seems like a great Open Source business as COBOL users are accustomed to paying for support and services. What do I not understand about this?
If cobol is so great, why do Amazon and google run exactly 0 lines of it?<p>There is a cost (vastly increased development time, inflexibility) to sticking with cobol, and it may be so great someday that it'll drive the companies that use it out of business. Just give newer companies 15 years after the current economic crisis is over and free competition and you'll see the difference.