Authors: Rodrigo Souza; Christina Chavez
Affiliation: Universidade Federal da Bahia
Abstract: Data from bug repositories have been used to enable inquiries about software product and process quality. Unfortunately, such repositories often contain inaccurate, inconsistent, or missing data, which can originate misleading results. In this paper, we investigate how well data from bug repositories support the discovery of details about the software verification process in two open source projects, Eclipse and NetBeans. We have been able do identify quality assurance teams in NetBeans and to detect a well-defined verification phase in Eclipse. A major challenge, however, was to identify the verification techniques used in the projects. Moreover, we found cases in which a large batch of bug fixes is simultaneously reported to be verified, although no software verification was actually done. Such mass verifications, if not acknowledged, threatens analyses that rely on information about software verification reported on bug repositories. Therefore, we recommend that the exploratory analyses presented in this paper precede inferences based on reported verifications.
Presented at the 9th Working Conference on Mining Software Repositories (MSR 2012).
Paper (PDF): msr2012-exploratory.pdf
Experimental package: msr2012-package.7z
Presentation video: http://www.youtube.com/watch?v=_jaJlvyLoTk