HOW DEVELOPERS TEST THEIR OPEN SOURCE SOFTWARE PRODUCTS - A Survey of Well-known OSS Projects

Abstract

Open Source Software (OSS) projects do not usually follow the traditional software engineering development paradigms found in textbooks, thus influencing the way OSS developers test their products. In this paper, we explore a set of 33 well-known OSS projects to identify how software quality assurance is performed under the OSS model. The survey investigates the main characteristics of the projects and common testing issues to understand whether a correlation exists between the complexity of the project and the quality of its testing activity. We compare the results obtained in our survey with the data collected in a previous survey by L. Zhao and S. Elbaum. Our results confirm that OSS is usually not validated enough and therefore its quality is not revealed enough. To reverse this negative trend, the paper suggests the use of a testing framework that can support most of the phases of a well-planned testing activity, and describes the use of Aspect Oriented Programming (AOP) to expose dynamic quality attributes of OSS projects.