Empirical evaluation of syntactic and semantic defects introduced by refactoring support

Software maintenance is a major source of expense in software projects. A proper evolution process is a critical ingredient in the cost-efficient development of high-quality software. A special case of software evolution is refactoring that cannot change the external behaviour of the software system yet should improve the internal structure of the code. Hence, there is always a need to verify after refactoring, whether it preserved behaviour or not. As formal approaches are hard to employ, unit tests are considered the only safety net available after refactoring. Refactoring may change the expected interface of the software therefore unit tests are also affected. The existing tools for refactoring do not adequately support unit test adaptation. Also, refactoring tools and guidelines may introduce semantic and syntactic errors in the code. This paper qualitatively and quantitatively analyses data from an empirical investigation involving 40 graduate students, performed against a set of semantic and syntactic defects. Findings from the expert survey on refactoring support have also been shared. The analysis in this paper shows that there are notable discrepancies between preferred and actual definitions of refactoring. However, continued research efforts are essential to provide Guide Lines (GL) in the adaptation of the refactoring process to take care of these discrepancies, thus improving the quality and efficiency of the software development.