Jan Tretmans – Model-Based Testing: There is Nothing More Practical than a Good Theory

Seminar: 25 April, 2017, 13:15, EF

Speaker: Jan Tretmans
TNO – Embedded Systems Innovation, Eindhoven, NL,
Radboud University, Nijmegen, NL,
Halmstad University, Halmstad, S.

We build ever larger and more complex software systems. Systematic testing plays an important role in assessing the quality of such systems. The effort for testing, however, turns out to grow even faster than the size and complexity of the systems under test themselves.One of the promising testing technologies to detect more bugs faster and cheaper is model-based testing. (more…)

Model-based testing starts with an abstract model of the system’s behaviour. This model constitutes a precise and concise specification of what the system shall do, and, consequently, is a good basis for the algorithmic generation of test cases and the analysis of test results. Model-based testing enables the next step in test automation by combining automatic test generation with test execution and providing more, longer, and more diversified test cases with less effort.

The presentation aims at covering the chain from theoretical concepts, via algorithms and tools, to industrial applications of model-based testing. Starting point is the ‘ioco’-testing theory for labelled transition systems, to which concepts from process algebra, the theoryof testing equivalences, symbolic transition systems, algebraic data types, satisfaction-modulo-theories tools, and equational reasoning are added. We show how these theories have led to the development of the model-based testing tool ‘TorXakis’.

On the one hand, TorXakis provides provably sound and exhaustive (in the limit) test generation from models. These models combine state-based control flow and complex data definitions, they deal with uncertainty through nondeterminism, they support compositionality by providing combinators for alternative, concurrent, sequential, exceptional, and interrupting behaviours, and they support abstraction and under-specification. On the other hand, TorXakis has shown practical usability in academia, both in research and in education,
as well as in industrial applications, ranging from smart-card software to large, systems-of-systems kind of applications. So, for model-based testing there is nothing more practical than a good theory.

Original announcement