Development of an efficient data coverage strategy for testmanager

Resumen

Model-based testing (MBT) emerged during the last years as a new approach to
deal with the testing of complex software systems, generating sets of test cases
automatically. MBT may be implemented following different approaches, the most
common ones being: transition, data and state coverage. This MSc thesis focuses
on the combination of transition and data coverage in order to achieve a more
complete testing. This combination is approached by the use of two black box
techniques: equivalence partitioning and boundary value analysis. GNU Prolog
(Gprolog) software, a constraint solver, has been selected as a solution for both
obtaining an immediate solution and achieving the domain of a constraint.
Achieved results on a small but realistic case study show acceptable execution
times in order to get 100% transition and boundary value coverage. Further results
are a boundary value coverage/time graph, an application of boundary value
analysis not only to numeric values but to most common data types, and a set of
metrics in order to evaluate results. All these achievements were gathered in a
boundary value strategy, named BVA, a data oriented strategy. As a conclusion, it
is shown that merging MBT approaches results in more benefits that the single use
of one of them.