Préface

Automated generation of tests from specifications has been recently introduced in order to address some of the shortcomings of classical automated unit testing. Test cases are created from properties that a piece of code must satisfy, and which are given by the programmer. Clearly this leads to much more concise test code. However, the aforementioned approach still has a long journey ahead before being fully ready to replace classical unit testing in most of its applications. Throughout this MSc Thesis we try to contribute towards the solution of the two main problems it is facing. First, test cases are currently generated from properties in a pure random fash-ion. We explore different alternatives for creating them in a more intelligent manner, but still very efficiently, so that the probability of finding bugs be-comes higher in practice. Second, properties are naturally suited to specify functional code, yet we need to develop new constructs to be able to apply them to stateful code, such as object-oriented software. We attempt to offer answers in this direction too. A prototype demonstrating our ideas for the resolution of both problems has been built as part of the Thesis. A snapshot of the source code at the time of writing can be found in the following URL. Further updates, if any, will also be made available on this site