Reflection on the Process of Open Sourcing Software Based on Ten Years of Development of RAPID

Abstract:

As the number and size of geoscience datasets persist in their constant rise, geoscientists increasingly express their yearning for further sharing of their data and software, and for facilitation of the associated academic credits. We present here our experience based on the open source development of an Earth System Model focusing on the propagation of water flow waves in large river networks: the Routing Application for Parallel computatIon of Discharge (RAPID). Since inception of RAPID ten years ago in January 2006, the community of its users has grown slowly but steadily, and now includes researchers in industry, academia, and government organizations. This growth of the RAPID users community can be explained – at least in part – by its open availability. However, despite an increasing support for open science (software and data), the mechanics of sharing still remain mysterious to many geoscientists… as they were for the authors. The purpose of this presentation is therefore to shed light on the steps involved in opening software and data based on a decade of experience related to the development and release of an Earth System Model. Three distinct steps of open sourcing are highlighted here: opening, exposing, and automatic testing. Each one of these steps is presented as an independent and tractable increment at various stages of development that is justified based on the size of the users community. Topics covered include software and data licenses, code and data repositories, unit testing, and continuous integration.