Résumé:
Companies are more and more forced to customize their software products for completely different customers. In practice they often clone an existing system and adapt it to the customer's needs. In such scenarios software product lines promise benefits, for example, reduced maintenance effort, improved quality, and customizability. However, introducing new development processes into a company is risky and might not pay off . The other advantage is that this fairly recent software development paradigm allows companies to create efficiently a variety of complicated products with a short lead-time. This thesis focuses on product planning and derivation, which is the process of creating individual products (members) during application engineering using shared product family artifact.
Firstly, we propose to apply Model-Driven Engineering techniques to provide a systematization of the Domain Engineering therefore to enable the automation of the Application Engineering. Model-driven techniques commonly rely on the use of metamodeling as a means to automate model-to-text and model-to-model transformations. In this work, we use ATL as a model-to-model transformation language and Acceleo as a model-to-text transformation language.
Secondly, due to the fact that a generalization relation may also exist between product lines. We propose a new structure to represent a composite software product line that allows members of the composition to communicate and interact with each other. The hierarchical structure proposed is based on inheritance that provides an easily understandable representation. The aim of the representation is to derive multiple products using a simple but practical method.
Finally, we illustrate the application of our approaches in various case studies in the context of e-Government Product Lines, from the feature model of each different product line to the final application. By applying the proposed approach, it becomes feasible to derive a number of applications in a specific domain
