An arc-based approach for visualization of code smells

Abstract

Code smells are indicators of design flaws that may have negative effects on software comprehensibility and changeability. In recent years several detection tools have been developed that are supposed to help in revealing code smells in large size software systems. However, usually a subset of the detected code smells are suitable for refactorings only. Previous studies on software clones have shown that visualization of findings may assist developers in identifying relevant refactoring opportunities by highlighting peculiarities and, thus, is useful to enhance a software's maintainability. Nevertheless, techniques to visualize code smells in general are rare, though, being an interesting field of research to bridge the gap between code smell detection and code smell refactoring. This paper presents a visualization approach that is supposed to help in assessing the dispersion and extent of arbitrary code smells by combining different existing techniques. The core of our approach consists of several Treemaps that are arranged on a circle in order to obtain a better integration of additional visualizations. Furthermore, the presented technique provides various interaction mechanisms that allow users to adjust the visualization to target elements of interest.