International Islamic University (IIUI), H10, Dept. of Computer Science, Islamabad, Pakistani.

---

Abstract—AutoInfer is a tool that is the state of the art in invariant detection. Invariants are properties of program components that remain unchanged throughout the execution of that component. AutoInfer automatically detects invariants for the programs under test which may or may not have a few invariant already present in them. AutoInfer uses AutoTest to generate a test suite for a given Program under test. The test suite (TS) is generated randomly using routine coverage as the coverage criteria. The generated TS is run on the program to create a change profile for the program, which in turn is used to activate relevant templates to generate quantified expressions. These expressions are the candidate contracts. Candidate contracts when run against the test cases are validated if they do not fail any test case. AutoInfer’s results are heavily based on the generated Test Suite. The better the generated Test Suite is better the final contracts will be. The work in this paper proposes that using Whole Test Suite (WTS) generation instead of random generation to generate the TS will results in a faster generation of the TS and will capture more errors as compared to AutoInfer. WTS is the state of the art in TS generation so it will result in better coverage and an improved fault detection capability.

Index Terms—Invariant detection, test oracles, software testing, AutoTest.

[PDF]

Cite: Sadia Ashraf, Almas Abbasi, "Invariant Detection Using Enhanced Autoinfer," Journal of Software vol. 12, no. 10, pp. 816-823, 2017.