Comparing SMartySPEM and vSPEM for Modeling Variability in Software Processes: A Qualitative Study

Abstract

Customizing and reusing software processes is a common practice for addressing the diversity of software projects. This fact expanded the number of Software Process Lines (SPrL) researches based on concepts derived from the Software Product Line approach. However, SPrL is still incipient with regard to existing literature providing de facto empirical studies on comparing SPrL variability modeling capabilities. In this context, approaches such as SMartySPEM and vSPEM, both based on the Metamodel Software and System Process Engineering Metamodel (SPEM), were proposed as annotative mechanisms for variabilities representation in SPrL. We compared SMartySPEM and vSPEM for comprehensibility and variability mechanism in a previous controlled quasi-experiment. Results from such study suggested further investigation on variability modeling of both approaches. Therefore, this paper presents an empirical study, which qualitatively compares such approaches based on Modularity, Traceability, Error Detection, Granularity, Adoption and Systematic Variability Management. Experts were invited to contribute to our study. Their feedback was analyzed based on Grounded Theory procedures, such as Coding. Results provided initial evidence SMartySPEM had better results compared to vSPEM, except for the Adoption criterion. In addition, several points of improvement for SMartySPEM were highlighted.