Status of Software Reuse 580

They usually do in most applications. There are relatively few large projects where the development of a specific language is part of the project, at least outside research projects. Most adapt some preexisting thing. I have worked on large projects where advanced statistical modeling was done in COBOL for arcane bureaucratic reasons.

I totally agree, but the problem is getting capital investors to believe in such developments up front. The circumstances which led to C's success were largely serendipitous. Multics was a failure by most critics. PL-I never really caught on. T(w)enex was a commercial blip. Even Ada is rarely used.

I don't think the current trend toward Java has anything to do with portability or reusability. It's one of those inexplicable fads. Now that WebSphere, et. al. have arisen around J2EE, it will roll over everything else, just as C did before, whether deserved or not. Like UNIX-C, web services based on J2EE will likely become the de facto standard for commercial development with little rational debate over relative merit.

An interesting issue is that the x86 architecture has become so universally dominant in the low to mid range that portability is becoming moot in many mainstream commercial applications.

I tend to agree with you about universal software reusability. Software is generally developed to meet some set of needs and if within those parameters it can be made extensible, so much the better. Otherwise, it is often a wasted effort, particularly in that it is difficult to anticipate what details will become important in the future.

I guess my point in starting this thread is that it seems that no matter how much software reuse has been ballyhooed, the overall state of things seems to me to have changed remarkably little over the past quarter century or more. Insightful programmers try to build it in, but the market selection pressure seems ambivalent at best.