Static analysis with its default configuration typically reports many issues, and some of them don't fit your projects. Most analysis tools have the capability to let users enable/disable analysis rules for users to avoid such "false positives." But, a huge amount of issues makes it difficult to manually verify all of them. For certain analysis tools listed in Supported Tools, Sider provides our recommended ruleset which can reduce false positives.
The recommended ruleset is obtained through a classification of coding-style rules using K-means machine learning algorithm. We've trained the classifier to classify which rules should be addressed or not based on the number of commits where the issue (coding-style violation) is reported and fixed. A total of about 3.6 million commits over 1,000 popular OSS repositories were used in this study.
This study has been conducted in collaboration with the following partners:
- Prof. Shingo TAKADA of Keio University in Kanagawa, Japan
- Assoc. Prof. Masataka NAGURA of Nanzan University in Aichi, Japan
Sider applies the recommended ruleset to the analysis tools below. Please follow the link for more details.