Towards Automated Restructuring of Object Oriented Systems

Software aging is an important cost contributor to the maintenance of aging software systems. Recent years have brought significant progress in the area of automatic detection of "code smells" as well as tool support for refactoring and implementing design patterns in the code. Nonetheless, there is hardly any tool support to help the maintainer decide how to refactor in a given situation, such that the recommended refactorings are also meaningful in that particular situation. Most of the existing techniques are either merely supporting the process, such as visualizations, or cannot guarantee meaningful refactorings, such as optimization based techniques. This paper introduces and experimentally evaluates a novel, tool supported approach to determine meaningful refactorings to structural flaws in object oriented systems. The refactorings recommended by our approach are guaranteed to lead to a meaningful and more maintainable structure in each analyzed situation. The approach contributes to a dramatic reduction of costs, by reducing the need and scope of detailed, manual code analysis