With this criteria, we tended to select successful projects of medium to large sizes, which cover an important subset of open source projects with defect-handling approaches and defect data that we can quantitatively evaluate. We can compare the results and findings to that of medium- and large-sized conventional software systems, which is the main object of studies for existing software engineering research.4 Limiting our focus to projects that meet these criteria also made our survey manageable, since publicizing the survey and handling correspondences might have been infeasible otherwise.