Guidance on Versioning of Digital Assets

Versioning of data and metadata is a crucial - but often overlooked - topic in scientific work. Using the wrong version of a (meta)data set can lead to drastically difference outcomes in interpretation, and lead to substantial, propagating downstream errors. At the same time, past versions of (meta)data sets are valuable records of the research process which should be preserved for transparency and complete reproducibility. Further, the final version of (meta)data sets may actually include errors that previous versions did not. Thus, careful version control is the foundation for trust in and broad reusability of research and operational (meta)data. This document provides an introduction to the principles of versioning, technical recommendations on how to manage version histories, and discusses some pitfalls and possible solutions. In the first part of this document, we present examples of change processes that require proper management and introduce popular versioning schemes. Finally, the document presents recommended practices for researchers as well as for infrastructure developers.