Towards an automation of the traceability of bugs from development logs: A study based on open source software

Abstract

Context: Information and tracking of defects can be severely incomplete in almost every Open Source project, resulting in a reduced traceability of defects into the development logs (i.e., version control commit logs). In particular, defect data often appears not in sync when considering what developers logged as their actions. Synchronizing or completing the missing data of the bug repositories, with the logs detailing the actions of developers, would benefit various branches of empirical software engineering research: prediction of software faults, software reliability, traceability, software quality, effort and cost estimation, bug prediction and bug fixing. Objective: To design a framework that automates the process of synchronizing and filling the gaps of the development logs and bug issue data for open source software projects. Method: We instantiate the framework with a sample of OSS projects from GitHub, and by parsing, linking and filling the gaps found in their bug issue data, and development logs. UML diagrams show the relevant modules that will be used to merge, link and connect the bug issue data with the development data. Results: Analysing a sample of over 300 OSS projects we observed that around 1/2 of bug-related data is present in either development logs or issue tracker logs: the rest of the data is missing from one or the other source. We designed an automated approach that fills the gaps of either source by making use of the available data, and we successfully mapped all the missing data of the analysed projects, when using one heuristics of annotating bugs. Other heuristics need to be investigated and implemented. Conclusion: In this paper a framework to synchronise the development logs and bug data used in empirical software engineering was designed to automatically fill the missing parts of development logs and bugs of issue data.