Software reuse has been widely believed to be one of the key approaches to relieve the chronic problems of software development. Reuse is defined as the reapplication of existing software entities in order to reduce the development cost of anew system [Bigg89]. Reusability, as another commonly cited notion in software engineering, is defined as a measure of the ease with which one can use those previously acquired entities [Prie87a]. There is a tradeoff between reuse and reusability. In order to increase the payoff from the reuse of a software component, we tend to make it very general in the sense that it can be used in as many cases as possible. The generality of reuse, however, has an adverse effect on reusability. A command that is capable of handling many cases usually has several arguments. The complexity resulting from such arguments degrades reusability. On the other hand, commands with relatively fewer arguments have high reusability but the payoff of reusing them is relatively low. To mitigate the tradeoff, we present an approach with specification by example to automatically set the arguments of executable programs (commands). In this approach, the user first locates the candidate command for reuse. Then the user specifies his/her intention by giving examples of the input and output. The system will automatically find a setting of the arguments by performing searching among the possible settings of the arguments. Revision of the input-output specification is allowed for fine-tuning input-output examples. When a proper setting for arguments is found, the user then can apply the program with the configured arguments to the intended data. A prototype has been implemented and several UNIX commands were successfully tested.
展开▼
