REUSABILITY OF MATHEMATICAL SOFTWARE - A CONTRIBUTION

Mathematical software is devoted to solve problems involving matrix computation and manipulation. The main problem limiting the reusability of existing mathematical software is the following: programs are often not initially designed for being reused. Therefore, it is hard to find programs that can be easily reused. In the first part or this paper, we give a programming methodology useful for designing and implementing reusable code. We name unit a portion of code designed and implemented for being reused. Our units are self-contained software components featuring a high degree of information hiding. This way of organizing software facilitates the reuse process and, furthermore, improves the understandability of units. To speed up the implementation process, a system supporting the reusability of units from an existing software library is particularly useful. In the second part of this paper, we report about an easy to use system of this kind.