Résumé
Reuse automation is a main concern of software engineering to produce high quality applications in a faster and cheaper manner. Some approaches define cross-platform model-driven software product lines to systematically and automatically reuse generic assets in software development. They improve the product line assets reusability by designing them according to the Model-Driven Architecture specifications. However, their reuse of platform-specific assets is limited due to an inefficient platform-specific variability management. This issue interfere with gains in productivity provided by reuse.In this paper, we define platform-specific variability by identifying variation points in different software concerns based on the well-known “4+1” view model categorization. Then, we fully manage platform-specific variability by structuring the Platform-Specific Model using two sub-models: the Cross-Cutting Model, obtained by transformation of the Platform-Independent Model; and the Application Structure Model, obtained by reuse of variable platform-specific assets. This structure is supported by a framework, based on a Domain-Specific Modeling Language, helping developers to build an application model. Experiments on three concrete applications confirmed that our approach significantly improves product lines productivity.