Preliminary Empirical Evidence on SPrL Variability Management with EPF and SMartySPEM

Abstract

Every day increases the level of demand by customers in regard to the quality and complexity of software systems. Because of this, companies are increasingly customizing their software processes according to market and project needs. A systematic way is the use of the Software Process Line strategy (SPrL), in which each product derived from the line is a specific software process. Therefore, variability management is an essential activity. This paper presents an empirical study comparing the compositional and the annotative approaches in representing variability in SPrLs taking into consideration a SCRUM-based SPrL. Eclipse Process Framework was chosen to represent the compositional approach, whereas SMartySPEM was considered to the annotative approach. The approaches were compared taking into account the sequential exploratory strategy based on mixed-methods. A qualitative empirical study comparing these approaches was planned, executed and previously published with relation to the following set of criteria: modularity, traceability, error detection, granularity and systematic management of variability. Such study was based on the expertise of software process experts and provided important information for the hypothesis formulation about systematic management of variability, the main dependent variable of this quantitative study. Thus, the quantitative study presented in this paper analyzes the effectiveness of variability representation. As the main contribution of this paper, we present preliminary evidence on the effectiveness of variability management, allowing supporting the gathering of solid evidence for further research in academic and industrial set about the compositional and annotative approaches for variability management in SPrLs. As a result of this quantitative empirical study it was not statistically possible to confirm that the annotative approach is more effective than the compositional approach.