In the August 1988 issue of the ACM SIGSOFT, Will Tracz, in his" Software Reuse Maxims" 1, argued (quoting M. Shaw's words) that Software Reuse will become the Expert Systems of th e 1990's. If it's true that, lately, Software Reuse has created a hype analogous to the one for Exper t Systems in the early 80's, then it's worth to wait and see whether early software reuse will experience the expert system shells craze of the late 80's.

Actually, the idea has been around for some time now2 and, definitely, early software reuse has bein g practiced since the first reusable software analysts, designers and engineers were invented. Reusing software at the earlier stages of its development, i. e. when it is still in the form of specifications, or designs is an intuitively appealing approach, mainly because of its big added value and its relativ e independence from implementation details. It is well known that requirements specification and design are probably the most labour-intensive, time-consuming, error-prone activities in softwar e development and therefore specifications and designs incorporate a significant added value. On the other hand, a specification and even a design will be ridden of all those implementation dependen t characteristics that make code so difficult to be reused. If we define the relative benefit from the reus e of software as