Reengineering of Component-Based Software Systems in the Presence of Design Deficiencies

The maintenance of component-based software systems requires up-to-date models of their concrete architecture, i.e. the architecture that is realised in the source code. These models help in systematically planning, analysing and executing typical reengineering activities. Often no or only outdated architectural models of such systems exist. Therefore, various reverse engineering methods have been developed which try to recover a system's components, subsystems and connectors. However, these reverse engineering methods are severely impacted by design deciencies in the system's code base, especially violations of the component encapsulation. As long as design deciencies are not considered in the reverse engineering process, they reduce the quality of the recovered component structures. Despite this impact of design deciencies, no existing architecture reconstruction approach explicitly integrates a systematic deciency detection and removal into the recovery process. Therefore, I have developed Archimetrix. Archimetrix is a tool-supported architecture reconstruction process. It enhances a clustering-based architecture recovery approach with an extensible, pattern-based deciency detection. After the detection of deciencies, Archimetrix supports the software architect in removing the de ciencies and provides the means to preview the architectural consequences of such a removal. I also provide a process to identify and formalise additional deciencies. I validated the approach on three case studies which show that Archimetrix is able to identify relevant deciencies and that the removal of these deciencies leads to an increased quality of the recovered architectures, i.e. they are closer to the corresponding conceptual architectures.