PROW: A Pairwise algorithm with constRaints, Order and Weight

• We present an algorithm called PROW for pairwise coverage.
• PROW algorithm allows constraint handling and prioritisation in pairwise coverage.
• We also present a web application called CTWeb that supports the PROW algorithm.
• CTWeb allows to obtain the set of products to test in Software Product Lines (SPLs), using a feature model as input.
• We present the application of CTWeb and PROW in a banking system and in a SPL of Sensors System.

Testing systems with many variables and/or values is often quite expensive due to the huge number of possible combinations to be tested. There are several criteria available to combine test data and produce scalable test suites. One of them is pairwise. With the pairwise criterion, each pair of values of any two parameters is included in at least one test case. Although this is a widely-used coverage criterion, two main characteristics improve considerably pairwise: constraints handling and prioritisation.This paper presents an algorithm and a tool. The algorithm (called PROW: Pairwise with constRaints, Order and Weight) handles constraints and prioritisation for pairwise coverage. The tool called CTWeb adds functionalities to execute PROW in different contexts, one of them is product sampling in Software Product Lines via importing feature models. Software Product Line (SPL) development is a recent paradigm, where a family of software systems is constructed by means of the reuse of a set of common functionalities and some variable functionalities. An essential artefact of a SPL is the feature model, which shows the features offered by the product line, jointly with the relationships (includes and excludes) among them. Pairwise testing could be used to obtain the product sampling to test in a SPL, using features as pairwise parameters. In this context, the constraint handling becomes essential. As a difference with respect to other tools, CTWeb does not require SAT solvers.This paper describes the PROW algorithm, also analysing its complexity and efficiency. The CTWeb tool is presented, including two examples of the PROW application to two real environments: the first corresponds to the migration of the subsystem of transactions processing of a credit card management system from AS400 to Oracle with .NET; the second applies both the algorithm and the tool to a SPL that monitors and controls some parameters of the load in trucks.