Title Apache Taverna: Sustaining research software at the Apache Software Foundation Authors Stian Soiland-Reyes http://orcid.org/0000-0001-9842-9718, Apache Taverna team Affiliation Apache Software Foundation; eScience Lab, University of Manchester Contact http://taverna.incubator.apache.org/contact URL http://taverna.incubator.apache.org/ License Apache Software License 2.0
Apache Taverna is a domain-independent scientific workflow system, encompassing a graphical workbench, command line tools, server and APIs. While Taverna was conceived for bioinformatics, its user base now also encase domains as diverse as astronomy, digital preservation, biodiversity and virtual physiology. Taverna has been developed open source since 2003 by the myGrid consortium, led by the University of Manchester and EMBL-EBI. In October 2014, Taverna became an incubating project at the Apache Software Foundation.
Here we describe our motivation for changing the governance and ownership of the Taverna project, and reflect on our experience and challenges of transitioning a University-led research software activity to an open development process and building a wider developer community.
Research Software supports scientists and researchers, its development usually funded for domain-specific projects and made publicly available as Open Source through code repositories like GitHub and BitBucket. When research software develop into a mature code base and gain a diverse user base, the initial funding and related projects may already be finished, yet ownership and control of the project typically remains with the original authors. Users and third-party developers have access to the source code, but are often not included in project decisions, and may not feel ownership to contribute code, documentation or support for others.
The original developers may eventually become involved in new projects that do not directly relate to the original project, and may loose focus on its changing user base. Thus it becomes critical to also grow a sustainable and diverse developer community, and to build an open governance model that encourages engagement and commitment from everyone using the software. Effort like the Software Sustainability Institute is crucial, as it helps guide research software engineers in best practices for making their research software open and maintainable, but ultimately, the success of an open source project should lead to a change of its structure and management to include anyone in its further development.
For Taverna, we considered several options to reduce the lead role and responsibility that University of Manchester had ended up with, and to move to a neutral ownership model where any interested developers could contribute to Taverna's development on an equal standing. One of the options was to create a Taverna Foundation, but this would have required legal administration and a dedicated budget.
Another option was to assign copyright and management of the code to a well-established software foundation like the Software Freedom Conservancy, the GNU project at the Free Software Foundation, the Eclipse Foundation, the Outercurve Foundation which is backed by Microsoft, and the Apache Software Foundation. The different options come with different implications for the project modus operandi, licensing, community, politics, infrastructure and public impression, which must be evaluated independently for different research software projects.
TODO: We chose Apache. Implications?
TODO: Results so far. Plans.