You are here:
Releasing Scientific Software in GitHub: A Case Study on SWMM2PEST
Citation:
Lin, X., M. Simon, AND N. Niu. Releasing Scientific Software in GitHub: A Case Study on SWMM2PEST. In Proceedings, SE4Science (ICSE), Montreal, Quebec, CANADA, May 28 - 31, 2019. Institute of Electrical and Electronics Engineers Incorporated (IEEE), Piscataway, NJ, 47-50, (2019). https://doi.org/10.1109/SE4Science.2019.00014
Impact/Purpose:
Releasing scientific software is a difficult, inflexible, and time consuming process. This study simplifies the release of scientific software via the GitHub process. Release engineering involves code development, integration, testing, and software delivery. It has been widely applied to deliver high-quality software to users. But while release engineering is a widespread practice in the software industry, there have been very few studies on the release engineering pipeline of scientific software. To fill this gap, we presented a study case in this paper to explain how the release engineering process is applied to the release of scientific software. We built a release workflow and released the parameter calibration software SWMM2PEST to GitHub. Moreover, we analyzed software version updates, software requirements changes, and the user forum to develop strategies for improving the ongoing release. The feasibility of improvement strategies has been proven by our consecutively released new versions of SWMM2PEST. The results offered insights for scientists to continuously release their scientific software.
Description:
Release engineering involves code development, integration, testing, and software delivery. It has been widely applied to deliver high-quality software to users. But while release engineering is a widespread practice in the software industry, there have been very few studies on the release engineering pipeline of scientific software. To fill this gap, we presented a study case in this paper to explain how the release engineering process is applied to the release of scientific software. We built a release workflow and released the parameter calibration software SWMM2PEST to GitHub. Moreover, we analyzed software version updates, software requirements changes, and the user forum to develop strategies for improving the ongoing release. The feasibility of improvement strategies has been proven by our consecutively released new versions of SWMM2PEST. The results offered insights for scientists to continuously release their scientific software.