Identifying Randomness related Flaky Tests through Divergence and Execution Tracing

Developers often spend time to determine whether test case failures are real failures or flaky. The flaky tests, known as non-deterministic tests, change their outcomes without any changes in the codebase, thus reducing the trust of developers during a software release as well as in the quality of a product. While rerunning test cases is a common approach, it is resource intensive, unreliable, and does not uncover the actual cause of test flakiness. Our paper evaluates an approach to identify randomness-related flaky. This paper used a divergence algorithm and execution tracing techniques to identify flaky tests, which resulted in the FlakyPy prototype. In addition, this paper discusses the cases where FlakyPy successfully identified the flaky test as well as those cases where FlakyPy failed. The papers discuss how the reporting mechanism of FlakyPy can help developers in identifying the root cause of randomness-related test flakiness. Thirty-two open-source projects were used in this. We concluded that FlakyPy can detect most of the randomness-related test flakiness. In addition, the reporting mechanism of FlakyPy reveals sufficient information about possible root causes of test flakiness.