Enhanced Code Coverage Approach for Regression Testing

Abstract

Code coverage information is widely used to address problems in software development and maintenance phase. Researchers have developed different approaches and tools using code coverage concept to accomplish regression testing tasks. These tools did not provide runtime enhancement techniques. In this research, an implementation of a proposed tool is conducted to measure the code coverage percentage which will be known as phase 1. Phase 2 focuses on enhancing the code coverage keeping the performance degradation as minimum as possible. A suite of selective case studies was tested to calculate the percentage of statement coverage, the number of executed statements and the overall runtime. The design of the proposed tool in this research is based on computing the overall runtime as a single thread approach. When it was tested on multiple threads a race condition problem was discovered as all the threads where trying to access the files in the same time which might lead to a deadlock. This implies resources wasting and delaying overall runtime. So the calculation of the overall runtime is assigned to a single highly configured processor to avoid transition time between multiple processors. This enhances the regression testing performance while decreasing the overall cost of the regression testing process. The proposed tool enhancement results are compared to its relative results in case of no enhancement process to judge the robustness of the proposed approach for this research. The comparison shows a range of enhancement fluctuation between 49%-67% for different sized test cases.