Systematic Testing of Graph Transformations: A Practical Approach Based on Graph Patterns

Abstract

Correctness is an essential property of model transformations. Although testing is a well-accepted method for assuring software quality in general, the properties of declarative transformation languages often prevent a direct application of testing strategies from imperative programming languages. A key challenge of transformation testing concerns limiting the testing effort by a good stop criterion. In this work, we tackle this issue for programmed graph transformations, and present a practical methodology to derive sufficient test suites based on a new coverage notion inspired by mutation analysis. We systematically generate requirement (graph) patterns from the transformation under test, applying different requirement construction strategies, and analyze the approach in terms of practicability, test suite quality and the ability to guide and support test case construction. © 2013 Springer-Verlag.