We started by identifying the projects of interest for our survey. Our first criterion was that the project had to have resulted in a product that the open source community generally recognized as a success. Our second criterion was that the project used its own publicly accessible defect database and defect-tracking tool. Because some projects, such as Apache or Jakarta, include many subprojects, we decided to include such subprojects as well because each subproject might handle defects somewhat differently. We also consulted individual project Web sites and used our prior
knowledge in selecting projects for our survey. This selection process yielded 75 open source projects or subprojects as our survey targets.