Multi-faceted Code Smell Detection at Scale using DesigniteJava 2.0
Tushar Sharma · 2024
Code smell detection tools not only help practitioners and researchers detect maintainability issues but also enable repository mining and empirical research involving code smells. However, current tools for detecting code smells exhibit notable shortcomings, such as limited coverage for a diverse kind of smells at varying granularities, lack of maintenance, and inadequate support for large-scale mining studies. To address the limitations, the first major version of DesigniteJava supported code smells detection at architecture, design, and implementation smells along with commonly used code quality metrics. This paper presents DesigniteJava 2.0 that adds testability and test smell detection support. Also, the tool offers new analysis modes, including an optimized multi-commit analysis mode, to support large-scale multi-commit analysis. We show that the optimized multi-commit mode reduces analysis time by up to 46% without compromising the analysis efficacy. The tool is available online. Replication package including all the validation data and scripts can be found online [27]. Demonstration video can be found on YouTube.