A RUP-Based Software Process Supporting Progressive Implementation

INTRODUCTION Software development has become a more complex activity over the last years. Clients have been increasingly demanding higher productivity, better software quality and shorter time to market. Additional strain results from new common requirements such as distribution and concurrent access. These and other non-functional aspects complicate implementation, test and maintenance activities. In order to simplify those activities, we argue that it is useful to tackle functional and non-functional concerns separately. In fact, whereas architectural and design activities should jointly consider both concerns [19], implementation activities can benefit from this separation. An implementation method might help programmers to effectively achieve this separation. Therefore, we have defined the progressive implementation method (Pim) [6], supporting a progressive approach for object-oriented implementation in Java [7], where persistence, distribution and concurrence control aspects are not initially considered in the implementation activities, but are gradually introduced. In this way we can significantly reduce the impact caused by requirements changes during development, and tame complexity by implementing and testing different aspects of code gradually. This progressive approach is possible because this method relies on the use of design patterns that provide a certain degree of modularity and separation of concerns [14], in such a way that the different aspects can be implemented separately. In this paper we extend the Rational Unified Process (RUP) [12] with Pim, providing proper implementation guidelines for RUP and hoping to support the progressive implementation of different aspects in software development projects where disciplined requirements, design and test activities are essential, demanding a software process. In Sections 2 and 3, we respectively present the main concepts of RUP and Pim, useful for a better understanding of our solution. Section 4 outlines the definition of RUPim, the proposed extension of RUP, and presents some results obtained in simple practical experiments using RUPim. Finally, Section 5 presents our conclusions and related work.