Investigation of Code Smells in Eclipse Framework Using SonarQube: An Empirical Analysis

Abstract

The Eclipse Framework provides both public and internal Application Programming Interfaces (APIs). Public APIs are widely supported and encouraged for use, while internal APIs are considered immature and subject to frequent changes. However, the quality of these APIs is not guaranteed, with many users reporting code smells, which could lead to application failures if unresolved. While some studies indicate that not all code smells can be easily fixed, users often face the challenge of either addressing these issues themselves or abandoning problematic APIs. To address this, we conducted an empirical investigation using the SonarQube static code analysis tool on 28 major Eclipse releases, aiming to identify code-smell-free APIs. Our study provides a dataset of 218K code-smell-free public APIs and 321K internal APIs. We found that 87.3% of public APIs and 91.5% of internal APIs in the analyzed releases are free from code smells, highlighting the importance of using these cleaner alternatives for application stability and long-term usability. Furthermore, we have discovered that the number of code smells proportionately increases as the Eclipse framework evolves. The average number of code smells and technical debt is 147K and 2,744 days respectively in all the studied Eclipse releases. Results from this study can be used by both interface providers and users as a starting point to recognize code smell-free interfaces and estimate efforts needed to fix code smells in each version of Eclipse.