Leveraging CVL to Manage Variability in Software Process Lines

Abstract

Abstract-Variability on project requirements often implies variability on software processes. To manage such variability, Software Process Lines (SPLs) can be used to represent commonality (i.e., common practices) and variability (i.e., differences) of a set of related software processes. To this end, some Software Process Modeling Languages (SPMLs) natively integrate variability mechanisms. Nevertheless, such a coupling between the SPML and the variability mechanisms i) requires to interpret the requirements variability in terms of the processes variability, ii) limits the reuse of the requirements variability for other purposes (e.g., the development itself), and iii) is a barrier to the use of advances from the field of variability management. In this paper, we propose an approach to apply the Common Variability Language (CVL from the OMG consortium) for requirement variability modeling and its binding to the processes. This work is illustrated on a family of industrial Java development processes. Our approach enables the definition of an SPL and the automatic derivation of a process from this SPL according to the requirements of a given project. The variability is managed separately from the process model and benefits from existing tools coming from process modeling community and CVL.

Integrating Software Process Reuse and Automation

“…Indeed, in a process line the processes are defined in intention, i.e., the parts common to several processes are factorized. In the field of software processes, several approaches rely on Software Process Line Engineering (SPLE) [7] to reuse software processes [2]- [6]. Among them, some approaches target SPEM 2.0 processes [2], [4], [5], while other approaches do not target a specific process metamodel [3], [6].…”

“…In a previous paper [6], we use CVL to define an SPL and to automatically derive a process from this SPL according to the requirements of a project. The gray part of Fig.…”

“…In the field of software processes, several approaches rely on Software Process Line Engineering (SPLE) [7] to reuse software processes [2]- [6]. Among them, some approaches target SPEM 2.0 processes [2], [4], [5], while other approaches do not target a specific process metamodel [3], [6]. In the field of business processes, that is closed to the field of software processes, there are also several approaches that rely on process line engineering in order to reuse business processes [16]- [26].…”

“…Therefore, existing approaches that aim at reusing software processes rely on the management of the variability of the software processes [2]- [6], using Software Process Line Engineering (SPLE) [7]. These approaches capture a family of software processes into a Software Process Line (SPL), that specifies commonalities and variabilities between these processes [3].…”

“…When the variability of a process under execution is not fully resolved, our approach consists of resolving this variability during the execution. To reuse the software processes from an SPL, we rely on a previous work [6] that consists of using the Common Variability Language (CVL) 1 [9] to manage the variability of software processes. To bridge the gap between the software processes and the components that automate them, we rely on another previous work [10] that consists of binding the components to the SPL in order to identify their contexts of use.…”