This repository contains a dataset of bug-fixing commits and the associated bug-inducing commits for seven different projects that is used in paper "Exploring and Exploiting the Correlations between Bug-Inducing and Bug-Fixing Commits". The bitex of this paper is as follows:
@inproceedings{wen2019exploring, title={Exploring and Exploiting the Correlations between Bug-Inducing and Bug-Fixing Commits.}, author={Wen, Ming and Wu, Rongxin and Liu, Yepang and Tian, Yongqiang and Xie, Xuan and Cheung, Shing-Chi and Su, Zhendong}, booktitle={Proceedings of the 2019 27th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering}, to appear, year={2019}, organization={ACM}}
This dataset is structured as follows:
./ACCUMULO.csv contains the information of project ACCUMULO. Each line contains the information of a bug, including bug-ID, bug-fixing commit and bug-inducing commit.
Be noted that a bug might contain multiple bug-inducing commits or bug-fixing commits.
Similar cases for the other projects AMBARI, HADOOP, JCR, LUCENE, and OOZIE.
./Defects4J.csv contains the information of bug-inducing commits of the Defects4J dataset (the 91 bugs whose bug-inducing commits have been identified through software testing).
./ACCUMULO: this folder contains the results of the project ACCUMULO, including the FileCoverage, LineCoverage, ActionCoverage and the results of SZZ. It also contains the information of our experimental data, including the code evolution between bug-inducing commits and bug-fixing commits.
Similar cases for the other projects AMBARI, HADOOP, JCR, LUCENE, and OOZIE.
./Statistical Testing of Previous Studies: this folder contains the results of revisting previous studies. Using R to run the script of test.R will generate the results of Table 2.
./Defects4J: this folder contains the results of the FL on Defects4J. Each subfolder in this folder contains the information of a bug. Specifically, ochiai.txt contains the results of SBFL, and induced.txt contains the results of the boosting model using the information of bug-inducing commits.
Be noted that when combining ochiai with the boosting model, remember to normalize the results of ochiai first.