Patterns for In-code Algebraic Testing

This paper describes an in-code approach to auto- matic algebraic-based software testing and a number of useful design patterns for doing it. The approach uses algebras as testable views on a system. These views form test interfaces which are highly automat- able. Specifications are expressed in terms of axioms of the algebras. We use the testing tool T2 to pro- vide automation. T2 works with in-code specifications; these are specifications written directly in a program- ming language. Because in-code specifications do not need any additional skill to master, they are more likely to be adopted by engineers on the field. Because they need no additional tools to parse and to keep them in- sync with the implementation, they are much cheaper to maintain. So, for real uses they have a good chance to scale up.