March 24, 2004

What programming languages are stable enough for long-term use?

Tim Bray talks about an interesting question, but I think he comes up with the wrong answer. He says "I think the Java answer is about as good as anyone’s at the moment", and maybe his recent overexposure to the Sun has affected his vision. Few of the corporate folks buying into Java are doing it from a long-term perspective; most are doing so because of the current coolness factor (well, current a few years ago).

One of the things that got people to trust long-term projects to COBOL 20-25 years ago was the fact that it was being standardized outside the vendor community. Sure, there were efforts by IBM to railroad the standards; fortunately for IBM's long-term sales, they failed. Java, on the other hand, is still tightly in the hands of Sun and no one expect Sun to do the Right Thing with it.

So, how does an IT director who is looking at decommissioning some COBOL-running mainframes pick a new language platform? They look at what has been fairly stable for ten years and seems fairly free of a single entity's control. Java isn't even an option; C, C++, and Perl 5 are. They all run on commodity hardware and commodity OSs (Linux and FreeBSD), they all play well with external database tools, and they are fairly easy to teach to the crop of new programmers who are being born today. It is likely that at least one of them will become as disliked 25 years from now as COBOL is today, and I predict that it will be C++: "too hard", "too arcane", and so on. Posted by lookit at March 24, 2004 05:25 PM