Computing Observable Modified Condition/Decision Coverage

by Daniela Grijincu for The Java Pathfinder Team

A significant issue when auto-generating tests is the effectiveness of the resulting test suite at finding faults. Structural coverage metrics, such as branch coverage, are used to guide automated test generation, but previous experiments have shown that such suites, when generated using model checking often do not provide good fault finding. This GSoC project will create an extension to JPF that will implement and evaluate automated test generation using a new test metric Observable Modified Condition/Decision Coverage (OMC/DC), which is an extension of the MC/DC metric that is used in safety-critical domains. Auto generated OMC/DC tests have been shown to be more effective at finding faults than MC/DC tests in a recent ICSE 2013 paper ("Observable Modified Condition/Decision Coverage" by Michael Whalen, Gregory Gay, Dongjiang You, Mats P. E. Heimdahl and Matt Staats) that targets data-flow programs (e.g. Simulink models). In this project, the notion of OMC/DC will be applied to Java files.