Software Product-Line Variability Management based on Model-Driven Engineering

A software product-line (SPL) is a family of products sharing the same assets. A product-line platform allows the derivation of distinct products within the same application domain. The use of SPLs appears to be the most effective strategy for achieving large-scale software reuse, assuring rapid development of new products with good quality levels. Enabling variability in software consists in delaying decisions at different software abstraction levels, ranging from requirements to runtime. A key issue in SPLs is its variability management, since it is the mechanism for derivation of specific products (SPL specialization). The mainstream object-oriented (OO) way to implement SPL variability is based on the development of frameworks reusable components of a software system described by a set of classes and the way instances of those classes collaborate. Model-Driven Engineering (MDE) is a trend which promotes the use of models as the main artifacts during software development and maintenance process. Models become artifacts to be maintained along with the code, by using model transformations and code generation. MDE is related with the Object Management Group (OMG) initiatives, Model-Driven Architecture (MDA) and Model-Driven Development (MDD), which argue that the use of models as the main artifact on software development will bring benefits on software reuse, documentation, maintenance, and development time. The aim of this PhD work is to provide an MDE-based technique for managing variability in SPLs covering different abstraction levels of variability in SPLs, assuming that its variability implementation is framework-based. The proposed approach intends to provide tool support to leverage difficulties in the SPL specialization process, by driving the application derivation gradually from requirements to implementation. Advantageous aspect-oriented technologies for separation of concerns in framework-based development, should be taken into account by this work. The potential main contributions of the proposed work are on SPL specialization comprehension, alignment between SPL representation and its implementation, structured and consistent SPL specialization documentation, efficiency of SPL specialization, modeling notations, and tool support for MDE-based SPL specialization. The technique should be validated by developing prototype CASE (Computer Aided Software Engineering) tool(s) implementing the technique, and performing a case study based on an industrial (or equivalent) real product-line system.