ValiPar: A testing tool for message-passing parallel programs

Abstract

The software testing activity is crucial for Software Quality Assessment. To aid at this phase, several testing criteria were proposed. A testing criterion is a predicate to he satisfied by a set of test cases. It is used to guide the selection and evaluation of a test data set and offers coverage metrics that quantify the testing activity. When parallel programs are considered, features as concurrency, communication and synchronization make more complex this activity. In this context. specific criteria and supporting tools are very important. This paper presents a tool called I a Ii Par, that implements testing criteria specific for parallel programs in message-passing environments. It provides a baseline to the selection and evaluation of test data. Based on the obtained coverage for a criterion the tester can evaluate the quality of the parallel program being tested.