Constraints: The core of product line engineering

Abstract

Abstract-Product line engineering is a reuse-driven development paradigm based on the management of variability, which was successfully applied in information systems engineering and other domains. A common way to represent variability is with variability models that describe artefacts, and the dependencies between their various inflexions. Constraint programming, and in particular Boolean constraint programming, has been used so far to support analysis of variability models such as Feature-Oriented Domain Analysis (FODA) and the like. This paper goes a step further by using constraint programming to specify product lines. The focus on variability, variation points or dependencies is switched to the concept of constraints that apply to variables. The paper shows that this approach is richer than the one based on dependencies. For instance, many constraints that were needed in the cases we explored cannot be specified with dependencies of existing product line modelling languages. The approach was implemented in a prototype tool, and its scalability explored with industry case studies. These experiments show that constraint programming encompasses existing product line modelling languages such as FODA or OVM (Orthogonal Variability Model) and opens way to new possibilities such as reasoning simultaneously with different models during domain or application engineering.

Model Driven Software Product Line Process for Service/Component-Based Applications

“…• Intention: it is a goal [6] that expresses what is targeted, i.e. a state that is expected to be reached or maintained.…”

Duplication Detection When Evolving Feature Models of Software Product Lines

“…For instance, [7] introduced the orthogonal variability model, which defines variability in a separate way. Salinesi et al [20] used a constraint-based product line language. Other approaches are proposed to model variability using UML models (Unified Modeling Language models) or feature models (FODA [3]).…”

Towards Parallel Constraint-Based Local Search with the X10 Language

“…On line 15, the solving process is executed and the returned cost is assigned to the cost variable. If this cost is equal to 0, the solver in a place has reached a valid solution, it is then necessary to send a termination signal to the remaining places (lines [16][17][18][19][20][21][22]. For this, every place (i.e.…”

“…Constraint Programming has been successfully used to model and solve many real-life problems in diverse areas such as planning, resource allocation, scheduling and product line modeling [16,17]. Classically constraint satisfaction problems (CSPs) may be solved exhaustively by complete methods which are able to find all solutions, and therefore determine whether any solutions exist.…”