Replication Package for article "diffDP: Using Data Dependencies and Properties in Difference Verification with Conditions"

This artifact provides the replication package for the article

Marie-Christine Jakobs, TU Darmstadt/LMU Munich, Germany
Tim Pollandt, TU Darmstadt, Germany
diffDP: Using Data Dependencies and Properties in Difference Verification with Conditions
In Proceedings of 18th International Conference on integrated Formal Methods (iFM), Springer, 2023.

The paper presents an approach that enhances difference verification with conditions. Difference verification with conditions is a regression verification technique. Its idea is to first extract an overapproximation of those paths in the modified
program that may cause property violations that do not exist in the original, unmodified software. In a second step, it restricts the verification of the modified program to the extracted paths. To this end, a difference detector extracts the overapproximation of the paths to be analyzed, which are then encoded into a condition. A condition is an automaton like structure known from conditional model checking to describe which paths of a program can be skipped and which need to be analyzed. The second step, then uses the condition and an arbitrary conditional model checker guided by the condition to restrict its analysis to the paths that need to be analyzed. When being proposed, one difference detector was suggested that works purely on syntactic differences and extracts all paths that are syntactically different, i.e., do not occur in the original program. The paper corresponding to this artifact introduces and evaluates a new difference detector, which is more complex and precise because it also takes data dependencies and program properties into account when extracting the relevant paths.

This artifact allows you to reproduce the experimental raw data of the experiments of the above paper and to regenerate all tables and figures presented in the paper from our raw data or your reproduced raw data. To this end, the artifact provides the software, benchmark programs, the benchmark files to reproduce the experimental raw data, our raw data, and the scripts we used to generate our tables and figures in the paper.

Our README provides detailed instructions on how to reproduce the results or a subset of it. In addition, we give a short tutorial on difference verification with conditions using the verifiers and extractors mentioned in the paper.

The artifact has been prepared to reproduce the paper's result with the iFM 2023 virtual machine.

Since the artifact contains software and data with different licenses, the artifact is not published under one common license. For more information about the licenses on the different components of the artifact, we refer to the LICENSE file provided with the artifact.