Taking a Studio Course in Distributed Software Engineering from a Large Local Cohort to a Small Global Cohort

Abstract

One of the challenges of global software engineering courses is to bring the practices and experience of large geographically distributed teams into the local and time-limited environment of a classroom. Over the last 6 years, an on-campus studio course for software engineering has been developed at the University of Queensland (UQ) that places small teams of students on different features of a common product. This creates two layers of collaboration, as students work within their teams on individual features, and the teams must interoperate with many other teams on the common product. The class uses continuous integration practices and predominantly asynchronous communication channels (Slack and GitHub) to facilitate this collaboration. The original goal of this design was to ensure that students would authentically experience issues associated with realistically sized software projects, and learn to apply appropriate software engineering and collaboration practices to overcome them, in a course without significant extra staffing. Data from the development logs showed that most commits take place outside synchronous class hours, and the project operates as a temporally distributed team even though the students are geographically co-located. Since 2015, a course adapted from this format has also been taught at the University of New England (UNE), an Australian regional university that is also a longstanding provider of distance education. In this course, most students study online, and the class has to be able to work globally, because as well as students taking part from around Australia, there are also typically a small number of students taking part from overseas. Transferring the course to a smaller but predominantly online institution has allowed us to evaluate the distributed nature of the course, by considering what aspects of the course needed to change to support students who are geographically distributed, and comparing how the two cohorts behave. This has produced an overall course design, to teach professional distributed software engineering practices, that is adaptable from large classes to small, and from local to global.