1. Model Based Testing Group 7  Nishanth Chandradas (1103346)  George Stavrinides (1128104)  Jeyhan Hizli (1002077)  Talvinder Judge (1113795)  Saajan Kothari (1138579)

2. Finite State Machines With Model Based Testing

3. While software testing, it is very useful to use models to make it easier by aiding test selection and test verification. Usually these models are just in the mind of the human tester who is also the developer. These mental models need to be written down in a shareable form such as UML. Finite state models are one of the most popular software models. We will discuss the advantages of using finite state machines for testing such as them being excellent because they describe the sequence of inputs that a software takes in addition to this they include many graph traversal algorithms. This makes it much easier to use them to test with opposed to manual testing. However we will also discuss the disadvantages such as the nontrivial construction and maintenance of complex software state machines.

4. What is a finite state machine?  It’s a mathematical model of computation  Used to design both computer programs and sequential logic circuits  Can be in one of a finite number of states

5. Advantages of using a finite state machine for software testing  Simple to create models of software  Predictable  Quick to design, implement and execute  Been around for a long time, allowing examples to learn from  Easy to transfer between abstract representation and coded implementation

6. Disadvantages of using a finite state machine for software testing  Can not be used in certain domains such as computer games.  Very difficult to manage complex and large finite state machines.  Sometimes over-specify implementation(sequencing is fully specified)  Number of states can be unmanageable  Numerical computations cannot be specified compactly (need extended FSMs)

7. Example of Finite State Machines Finite number of states (Blue) Triggers (Orange) To move between states a trigger must be activated When testing, if a trigger is activated, and the appropriate state is not met, then it is clear there is an error within the software being tested.

8. Labelled transition systems and Ioco testing theory.  In model based testing: A model is tested for compliance and its behaviour is tested.  Labelled transition systems is this “model” and “compliance” defined with the Ioco testing theory.

9. Labelled transition systems  A structure consisting of states with transitions, labelled with actions between them. Transition State Action

10. Labelled transition systems  Introduced as models for specifications, implementations and tests.  Defined as a process language for representing complex transition systems.

11. IOCO testing theory  Formal specification: What the implementation should do or not do. In formal, expressed as language to see if behaviour conforms to spec.  Test cases : Conditions the implementation should pass to comply with the correct behaviour.  Implementation relation ioco : defines conformance between implementations and specifications. “Expressing correctness”  Test generation algorithm : Derives a set of tests from spec. Check whether IMP IOCO conforms to that spec.

12. PROS  Provides a sound and well-defined foundation for labelled transitional systems.  Practical basis for several model based test generation tools and applications. Cons  Theory is only useful if it is supported by model based test tools, in particular generation tools.

13. Examples  LTS for coffee machine implementation under test: Input of £1 makes coffee Input of 50p makes tea Machine returns £1 if any error