Improving source code change set analysis by using a visual tool. Controlled experiment dataset.

Modern software development is performed by developing features in isolated branches by each member of a software development team. When these branches need to be integrated, they have to be manually reviewed by an integrator. Source code reviewing can be a tedious and time consuming task, which is normally performed by hand reviewing a textual diff of the change set.

The difficulty of reviewing source code changes can have a negative impact on the accep- tance of these changes. It has been shown that this can imply the rejection of important bug fixes for a software project.

In order to facilitate the task of reviewing source code change sets, we designed and implemented a visual tool. Our visual tool helps in assessing source code change sets by providing different views of the change set: an overall overview of the change set with metrics, and a visualization representing the structural changes in the source code. We evaluated our visual tool by performing a controlled experiment with software practitioners.

Our experimental results show no significant differences between using our visual tool and a textual diff tool in the following terms: time of analysis, precision of the analysis, and inference of intention of changes. However, we did find a positive user perception, and a reduced mental load when using our visual tool.

Anonymized datasets with the answers to the controlled experiments for evaluating the Git Thermite visual tool for assessing source code change sets. Copies of these datasets are provided in both, CSV format, and OpenDocument format.

Exact copies of the learning materials that was provided to the participants during the execution of the controlled experiments is also included along the dataset.