A deductive reasoning approach for database applications using verification conditions

Deductive verification has gained paramount attention from both academia and industry. Although intensive research in this direction covers almost all mainstream languages, the research community has paid little attention to the verification of database applications. This paper proposes a comprehensive set of Verification Conditions (VCs) generation techniques from database programs, adapting Symbolic Execution, Conditional Normal Form, and Weakest Precondition. The validity checking of the generated VCs for a database program determines its correctness w.r.t. the annotated database properties. The developed prototype DBverify based on our theoretical foundation allows us to instantiate VC generation from PL/SQL codes, yielding to detailed performance analysis of the three approaches under different circumstances. With respect to the literature, the proposed approach shows its competence to support crucial SQL features (aggregate functions, nested queries, NULL values, and set operations) and the embedding of SQL codes within a host imperative language. For the chosen set of benchmark PL/SQL codes annotated with relevant properties of interest, our experiment shows that only 38% of procedures are correct, while 62% violate either all or part of the annotated properties. The primary cause for the latter case is mostly due to the acceptance of runtime inputs in SQL statements without proper checking.We would like to thank Mr. Moolchandra Mridul and Mr. Nabeel Qaiser for their help in the development of DBverify. We also thank the anonymous reviewers, Area Editor, and Editor in-Chief of the journal for their valuable comments and helpful suggestions. This work is partially supported by the IMPRINT-2 Project (IMP/2018/000523) from the Science and Engineering Re search Board (SERB), Department of Science and Technology, Government of India and by National Funds through the Portuguese funding agency, FCT - Fundação para a Ciência e a Tecnologia within project UIDB/50014/2020