Our paper investigates the reproducibility of software defect datasets. Specifically, we present a study on the reproducibility of five Java software defect datasets, examine the reproducibility of one of them in a 13-month period, propose fixes for software breakages, and explore two ways to reduce breakage for long-term reproducibility.

This artifact consists of two parts: raw data used in the study, and instructions on how to replicate the study’s findings using the original raw data, and/or from scratch. We provide the main steps to be replicated by the Artifact Evaluation Committee members, along with links to the full documentation for each step, which can be found in our public repository: https://github.com/ucd-plse/On-the-Reproducibility.

Please see the abstract PDF attached for more information.