Software systems have become ubiquitous over the last twenty years. However, software is not only present in every-day life, it has also become extremely important in all fields of science as well. Software, and in particular Data Analysis Workflows (DAWs for short), is thus becoming even more important than ever. Unfortunately, in many scientific fields these massively complex workflows have to be developed, implemented, and maintained by the domain scientists themselves, who by and large have never received a formal software engineering education. While software engineering principles significantly advanced our capabilities to create complex software-intensive systems, designing software that meets specific requirements, the resulting DAWs in the scientific fields oftentimes suffer from the same exact problems we as software engineers saw during the so-called software crisis: architecture degradation, software projects growing uncontrollably, loss of maintainability, to name just a few. It is an open research question on how to test and debug such systems in an automated (or at least semi-automated) fashion.

Creative Commons BY 4.0

Sebastian Müller