Investigating the Relation between Code Comments and Issue Resolution Time in Java Repositories

Source code comments play a critical role in software development by facilitating issue resolution and reducing maintenance time and cost. However, the impact of code comments, particularly those aiding program comprehension, on expediting issue resolution remains insufficiently explored. This study investigates the impact of code comments on issue resolution time for issues related to pulling requests involving code maintenance. To this end, ten of the most popular Java repositories on GitHub were selected, and closed pull request-related issues were identified using a specific criterion. Comments were then extracted from classes modified by the pull requests, and a trained BERT model was utilized to classify them according to their usefulness. The percentage of comments relative to code size, the percentage of useful comments relative to all comments, and the average issue resolution time were computed and examined for correlations. The results indicate a negative correlation (-0.68) between the percentage of code comments in modified classes and the average issue resolution time. A stronger negative correlation (-0.73) was found between the percentage of useful code comments to all comments in modified classes and the average issue resolution time per class. These findings demonstrate that code with more comments, particularly useful ones, leads to faster issue resolution. This serves as a basis for further research into the impact of various fine-grained comment types on issue resolution time in different programming languages. Also, it can assist developers in writing more effective comments that accelerate bug-fixing and reduce software maintenance efforts.