Detecting and Classifying Nonconformances in Code Contracts with ContractOK

Abstract

Nonconformances, in the context of contract-based programs, must be detected and corrected. Classification may be useful in the process of nonconformances correction. Current approaches do not support any type of nonconformance classification. In this work, we present a dynamic approach (CONTRACTOK) for detecting and classifying nonconformances in the context of Code Contracts programs. The approach is based on random test generation for nonconformances detection and on heuristics for classification. We evaluate our approach in four real programs, summing up 82.8K lines of C# and Code Contracts, detecting and classifying 16 nonconformances.