Exploring the communication functions of comments during bug fixing in Open Source Software projects

Sandra L. Ramirez-Mora, Hanna Oktaba, Helena Gómez-Adorno, Gerardo Sierra

Abstract: Context: Bug fixing is a frequent and important task in Open Source Software (OSS) development and involves the communication of messages, which can serve for multiple purposes and affect the efficiency and effectiveness of corrective software activities. Objective: This work is aimed at studying the communication functions of bug comments and their associations with fast and complete bug fixing in OSS development. Method: Over 500K comments and 89K bugs of 100 OSS projects were extracted from three Issue Tracking Systems. Six thousand comments were manually tagged to create a corpus of communication functions. The extracted comments were automatically tagged using machine learning algorithms and the corpus of communication functions. Statistical and correlation analyses were performed and the most frequent comments communicated during fast and successful bug fixing were identified. Results: Significant differences in the distribution of comments of fixed and not fixed bugs were found. Variations in the distribution of comments of bugs with different fixing time were also found. Referential comments that provided objective information were found to be the most frequent messages. Results showed that the percentages of conative and emotive comments are greater when bugs are resolved without the requested fixes and when fixes are implemented in a long time. Conclusion: Associations between communication functions and bug fixing exist. The results of this work could be used to improve corrective tasks in OSS development and some other specific linguistic aspects should be studied in detail in OSS communities.