A Classification Model for Reusable Software Components

Abstract

TIris paper presents work which has been carried out in the ESF-ROSE project (referred to as ROSE in the remainder of this paper). Funded under the Eureka programme (Eureka is the famous interjection used by Archimedes in his bath and not an acronym), the ESF project (Eureka Software Factory) aims at providing a highly effective software manufacturing environment The use of the wordfactory in the software context has very little connection to traditional assemblyline factories. Instead, a Software Factory is a factory in the modem sense providing Computer Integrated Software Manufacturing with emphasis on integration. A software factory covers the total software production process, including all technical and managerial tasks, with a high degree of automation and resource utilization. The ROSE project is a collaborative effon involving MATRA Espace, the software house Serna Group (France), and the University of Dortmund (Federal Republic of Gennany). Pan of the work on classification is the Ph.D. research of author Eric Rames. The main goals of the ROSE project (Reuse Of SoftwarE) are: • to analyze and to define in a comprehensive way the concept of software reuse; • to develop an environment for the reuse of software components within a factory. A precondition for reuse in software development is the existence of libraries of reusable software components. In order to suppon reuse, the collection must contain not only the components themselves, but also be accessible by a system that provides descriptions of the components and retrieval mechanisms so that users may match their specific requirements against these descriptions. Indexing reusable software components according to a classification scheme allows reusers to have a better understanding and more efficient access to the libraries' contents. Therefore a classification scheme is built so that it represents selection criteria the reuser might have. These indexes would be searchable and keyed to retrievable software descriptions. Based on retrievals, users may then access the actual software. Building such a collection is a domain analysis process [PRI90] that includes activities such as: • Identification of software components that should be reusable and description in terms of reusable software components. • Definition of a classification scheme appropriate for indexing and retrieving the reusable software components. TIris paper focuses on the latter topic and how it is performed in the ROSE project. A case study carried out from the aerospace domain is then presented. A discussion of ongoing and future work will conclude this paper.