Search

It is a long-standing hypothesis that the concise and customized notation of a DSL improves the performance of developers when compared with a GPL. For non-technical domains—e.g., science—, this hypothesis lacks empirical evidence. Given this lack of empirical evidence, we evaluate a DSL for ecological modeling designed and implemented by us with regard to performance improvements of developers as compared to a GPL. We conduct an online survey with embedded controlled experiments among ecologists to assess the correctness and time spent of the participants when using a DSL for ecosystem simulation specifications compared with a GPL-based solution. We observe that (1) solving tasks with the DSL, the participants’ correctness point score was —depending on the task— on average 61 % up to 63 % higher than with the GPL-based solution and their average time spent per task was reduced by 31 % up to 56 %; (2) the participants subjectively find it easier to work with the DSL, and (3) more than 90 % of the subjects are able to carry out basic maintenance tasks concerning the infrastructure of the DSL used in our treatment, which is based on another internal DSL embedded into Java. The tasks of our experiments are simplified and our web-based editor components do not offer full IDE-support. Our findings indicate that the development of further DSL for the specific needs of the ecological modeling community should be a worthwhile investment to increase its members’ productivity and to enhance the reliability of their scientific results.

Despite the increasing importance of in silico experiments to the scientific discovery process, state-of-the-art software engineering practices are rarely adopted in computational science. To understand the underlying causes for this situation and to identify ways to improve it, the authors conducted a literature survey on software engineering practices in computational science. They identified 13 recurring key characteristics of scientific software development that are the result of the nature of scientific challenges, the limitations of computers, and the cultural environment of scientific software development. Their findings allow them to point out shortcomings of existing approaches for bridging the gap between software engineering and computational science and to provide an outlook on promising research directions that could contribute to improving the current situation.

To adapt established methods of software engineering for scientific computing, we propose a software development approach for interdisciplinary teams of scientists called Sprat. The approach is organized around a hierarchical architecture that combines internal and external domain-specific languages (DSLs). For its evaluation, Sprat is employed in the implementation of a marine ecosystem model. We highlight what is to be observed while integrating the DSLs into the hierarchy in order to enable a successful cooperation of scientists in interdisciplinary teams as well as to achieve a maintainable code base.