Coordinating software reuse activities is a complex problem when considering collaborative software development. This is mainly motivated due to the difficulty in specifying how the artifacts and the knowledge produced in previous projects can be applied in future ones. In addition, modern software systems are developed in group working in separate geographical locations. Therefore, techniques to enrich collaboration on software development are important to improve quality and reduce costs. Unfortunately, the current literature fails to address this problem by overlooking existing reuse techniques. There are many reuse approaches proposed in academia and industry, including Framework Instantiation, Software Product Line, Transformation Chains, and Staged Configuration. But, the current approaches do not support the representation and implementation of collaborative instantiations that involve individual and group roles, the simultaneous performance of multiple activities, restrictions related to concurrency and synchronization of activities, and allocation of activities to reuse actors as a coordination mechanism. These limitations are the main reasons why the Reuse Description Language (RDL) is unable to promote collaborative reuse, i.e., those related to reuse activities in collaborative software development. To overcome these shortcomings, this work, therefore, proposes CollabRDL, a language to coordinate collaborative reuse by providing essential concepts and constructs for allowing group-based reuse activities. For this purpose, we extend RDL by introducing three new commands, including role, parallel, and doparallel. To evaluate CollabRDL we have conducted a case study in which developer groups performed reuse activities collaboratively to instantiate a mainstream Java framework. The results indicated that CollabRDL was able to represent critical workflow patterns, including parallel split pattern, synchronization pattern, multiple-choice pattern, role-based distribution pattern, and multiple instances with decision at runtime. Overall, we believe that the provision of a new language that supports group-based activities in framework instantiation can help enable software organizations to document their coordinated efforts and achieve the benefits of software mass customization with significantly less development time and effort.

Journal of Systems and Software