Recommendation of Refactorings for Improving Dependability Attributes

Abstract

The incidence of design problems (DP) in systems is associated with the quality decay of dependability attributes such as maintainability and robustness. This often results in negative consequences such as rework and error proneness. Thus, there are many techniques for supporting the resolution of DPs through refactoring recommendations. However, in practice, deciding where and when to refactor is still a challenging task. In addition, existing techniques fail to provide effective support. Thus, we aim at advancing the state-of-the-art through a flexible refactoring recommendation technique. Our technique is intended to help practitioners in improving multiple dependability attributes. We hypothesize that our recommendation technique will be able to overcome existing limitations by using empirically validated heuristics. Our heuristics will use information such as the density, diversity, and granularity of symptoms to recommend refactoring opportunities. The effectiveness of our technique will be evaluated through empirical studies involving heterogeneous software projects and experienced software practitioners.