Semi-Automated Test-Case Propagation in Fork Ecosystems

Abstract

Forking provides a flexible and low-cost strategy for developers to adapt an existing project to new requirements, for instance, when addressing different market segments, hardware constraints, or runtime environments. Then, small ecosystems of forked projects are formed, with each project in the ecosystem maintained by a separate team or organization. The software quality of projects in fork ecosystems varies with the resources available as well as team experience, and expertise, especially when the forked projects are maintained independently by teams that are unaware of the evolution of other's forks. Consequently, the quality of forked projects could be improved by reusing test cases as well as code, thereby leveraging community expertise and experience, and commonalities between the projects. We propose a novel technique for recommending and propagating test cases across forked projects. We motivate our idea with a pre-study we conducted to investigate the extent to which test cases are shared or can potentially be reused in a fork ecosystem. We also present the theoretical and practical implications underpinning the proposed idea, together with a research agenda.