An Investigation of the Fault-proneness of Clone Evolutionary Patterns

Requirements

• Python 2.7 or newer
• R 3.1 or newer
• NiCad clone detection tool
• iClones clone detection tool

File description

• analytic_scripts folder contains the scripts to mine software repositories.
  • detect_clones.py: detect clone classes from each of the commits from a Git repository. This script will output raw clone results of a subject system.
  • extract_clone_results.py: extract clone classes into a JSON file from the raw clone results.
  • build_genealogies.py: extract clone pairs from the results of clone classes, then built the clone genealogy for each clone pair.
  • analyse_genealogies.py: extract basic metrics from the results of clone genalogies.
  • independant_variables.py: extract explanatory metrics from the basic metrics of clone genealogies.
  • commit_bug_mapping.py: map bug-fixing commits to their corresponding bugs.
  • fault_inducing.py: identify bug-inducing commits based on the SZZ algorithm.
• statistics folder contains data extracted from software repositories and statistical analysis scripts. For each subject system, we extract its basic metrics; and from the basic metrics we further extract its metrics for modelling.
  • modelling folder contains R scripts to build the GLM model for each subject system using a specific clone detection tool.
• raw_data folder contains the commit logs and commit sequence for each subject system extracted from its Git repository.
• output_data.zip is a compressed folder that contains the results of clone classes, clone genealogies, and bug-inducing commits.

How to user the analytic scripts

1. Clone a project's Git repository. For systems originally managed by SVN, please follow this tutorial to clone the repository as Git: https://www.atlassian.com/git/tutorials/migrating-convert/.
2. Use the following command to extract the project's commit logs: git log --pretty=format:"%H,%ae,%ai,%s".
3. Uncompress the folder output_data.zip.
4. Run detect_clones.py and extract_clone_results.py to detect clone classes for a subject system using a clone detection tool.
5. Run build_genealogies.py to extract clone pairs from the JSON file, then build clone genealogies for each clone pair.
6. Run commit_bug_mapping.py and fault_inducing.py to identify bug-inducing commits.
7. Run statistics.py to perform Fisher's exact test for RQ1 and RQ2.
8. Run independant_variables.py to extract explanatory variables for RQ3.
9. Build GLM models with the R script in the statistics/modelling folder for RQ3.

Data source

• Source code repositories:
  • ArgoUML: http://argouml-stats.stage.tigris.org/documentation/defaulthtml/cookbook/ch02s05.html
  • Ant: https://github.com/apache/ant
  • JEdit: https://sourceforge.net/p/jedit/svn/HEAD/tree/
  • Maven: https://github.com/apache/maven
• Bug tracking systems:
  • ArgoUML: http://argouml.tigris.org/project_bugs.html
  • Ant: https://bz.apache.org/bugzilla/
  • JEdit: https://sourceforge.net/p/jedit/bugs/
  • Maven: https://issues.apache.org/jira/browse/MNG

For any questions

Please send email to le.an@polymtl.ca