Effects of Version Control System on the Efficiency of Refused Bequest Code Smell Identification Algorithm

Refused bequest is a code smell that originates from introduction of inappropriate class hierarchies where subclasses do not specialize or use the functionalities provided by the parent class. The existing refused bequest code smell identification algorithms use either static analysis or a combination of static and dynamic analysis based technique. However, those algorithms do not consider version control history. If the parts that have been changed from a previous version can be considered for refused bequest identification, the identification time may be reduced. An approach is proposed here employing the version control history to reduce identification time in refused bequest code smell identification. The proposed approach has four components namely Change Collector, Source Code Parser, Error Injector and Smell Calculator. Change Collector identifies source codes that have been changed from the previous version. Source Code Parser then parses the source codes to identify metrics such as inherited method, overridden method, super classs method invocation etc. After parsing the source codes, Error Injector introduces intentional errors in the non-overridden inherited methods of the subclasses. Smell Calculator then calculates smell for each of the subclasses based on metrics identified in Source Code Parser and Error Injector. The comparative analysis of the results shows a significant improvement in execution time. The proposed approach is found to be 16.2% more time efficient than the existing approach for the sample projects. This is because the approach incorporates version control history in the identification process. Moreover, the proposed approach is able to identify code smells as accurately as the existing approach.