Master Thesis MSTR-2021-42

BibliographyGrotepass, Florian Frank: Analyzing the quality of RESTful APIs in a migration context: a case study in the e-commerce domain.
University of Stuttgart, Faculty of Computer Science, Electrical Engineering, and Information Technology, Master Thesis No. 42 (2021).
85 pages, english.

In this work, we compare two RESTful API versions with respect to their software product quality. These APIs are maintained by the Cologne-based company Trusted Shops. Both API versions provide similar functionality. The company wants to replace the old API version with the new version and ensure that the software product quality has improved. Therefore, we conducted a three-step case study where we first interviewed both API users and maintainers. We then conducted a qualitative data analysis of these interviews. Second, we manually detected (anti)-patterns and violations of common design rules and best practice in both API specifications. Lastly, we evaluated the two specifications with the help of four different analysis tools, three rule-based analysis tools and one tool based on quality metrics. All in all, the new API version eTrusted was perceived as superior over the old API version myTS by both API maintainers and users. However, only the manual search for (anti-)patterns and violations of API rules and best practices came to a similar conclusion as the qualitative data analysis. We identified eleven violations for myTS and only ten for eTrusted. The tool-supported evaluations showed different results. We found only 18 rule violations for the old API version but 19 for the new API version with the help of rule-based tool analysis. Also, seven out of ten metrics reported a better result for the old API version. Therefore, rule-based analysis and metric-based evaluation cannot be taken as an absolute indicator for software product quality. However, these techniques are still powerful to find low-level flaws in API specifications. Furthermore, with the help of metric-based evaluation, specific partial aspects of quality attributes can be determined. However, more research towards a uniform ruleset of service-interface (anti-)patterns and best practices is needed. Also, some metrics have to be validated.

Full text and
other links
Department(s)University of Stuttgart, Institute of Software Technology, Empirical Software Engineering
Superviser(s)Wagner, Prof. Stefan; Bogner, Dr. Justus
Entry dateNovember 4, 2021
   Publ. Computer Science