Software Reusability through Versions

Abstract

In this paper a new approach for modelling program variants (versions) is proposed, which is focused on increasing the level of software reuse, rather than on enriching data model. In this approach, a program is composed of a program body and a set of logically independent program contexts. The program body contains global functions and global data structures. Each program context contains exactly one variant of every class defined in the program and one variant of every non-global function, called a context function. Variants of the same class/function belonging to different contexts need not be different. During the program execution only one context is active. It may, however, be changed dynamically at program run-time. Thus, at a particular moment of time the program is viewed as a sum of the program body and exactly one program context.