ITR: Collaborative Research: A New Generation of Scalable, Cost-Effective Regression Testing Techniques

CCR-0080898 PI：塞巴斯蒂安·埃尔鲍姆大学。内布拉斯加州 - 林肯 ITR:SW：协作研究：新一代可扩展、经济高效的回归测试技术 合作 PI：Gregg Rothermel (CCR-0080900) 该项目是与俄勒冈州立大学的 Gregg Rothermel (CCR-0080900) 合作开展的。回归测试是在修改后的软件上执行的一个昂贵的过程，以确保修改不会损害其质量。 为了帮助完成这个过程，之前的研究已经考虑了各种测试套件重用技术。 尽管这些技术取得了进展，但它们在几个方面仍然受到限制。 例如，他们通常假设测试用例具有相同的成本，故障具有相同的严重性，并且故障可能性在程序的各个部分中是恒定的。 这些假设在实践中是不现实的，并且限制了技术的适用性和有效性。 拟议的研究将解决这些局限性。 该研究将提供：（1）捕获必要因素的综合回归测试成本模型； (2) 考虑这些因素的回归测试技术； (3) 更准确地理解这些因素对回归可测试性、程序和测试设计以及软件工程实践的影响； (4) 帮助软件工程师选择和创建具有成本效益的回归测试工具和流程的指南。 除了提供模型、算法和流程之外，该研究还包括大量的实证成分，并将提供有关技术和因素的公开可用的实证数据基础。 这些贡献将共同支持更高效、更有效的回归测试，并提高软件质量。

CCR-0080898 PI: Sebastian Elbaum Univ. of Nebraska - Lincoln ITR:SW: Collaborative Research: A New Generation of Scalable, Cost-Effective Regression Testing Techniques Collaborating PI: Gregg Rothermel (CCR-0080900)This project is carried out in collaboration with Gregg Rothermel (CCR-0080900)of Oregon State University. Regression testing is an expensive process performed on modified software to provide confidence that modifications have not impaired its quality. To help with this process, previous research has considered various test-suite reuse techniques. Despite progress with these techniques, they remain limited along several dimensions. For example, they typically assume that test cases have equivalent costs, faults have equivalent severities, and fault likelihood is constant across portions of a program. These assumptions are unrealistic in practice, and limit the applicability and effectiveness of techniques. The proposed research will address these limitations. The research will provide: (1) comprehensive regression testing cost models that capture the necessary factors; (2) regression testing techniques that account for these factors; (3) more precise understanding of the effects these factors have on regression testability, program and test design, and software engineering practice; and(4) guidelines that help software engineers select and create cost-effective regression testing tools and processes. In addition to providing models, algorithms, and processes, the research includes a substantial empirical component, and will provide a publicly available base of empirical data about techniques and factors. Together, these contributions will support more efficient and effective regression testing, and improve the quality of software.