Scalable Program-Analysis-Based Testing and Maintenance: Infrastructure and Experimentation

软件的维护和测试费用很高。程序分析技术分析源代码以获得有关软件的信息，它承诺通过提供信息来帮助软件工程师执行特定的维护和测试任务，如程序理解、影响分析和回归测试，从而降低这一费用。然而，要被采用，甚至被考虑采用，基于程序分析的测试和维护技术必须经过经验验证，才能扩展到大型软件系统。这项研究的主要目标是经验性地调查基于程序分析的测试和维护技术对大型系统的适用性。这项研究涉及三个总体努力：(1)构建一个可扩展的实验基础设施，包括软件系统和主题系统的知识库；(2)开发支持大规模系统测试和维护的程序分析技术；(3)对基础设施以及基于程序分析的测试和维护技术进行严格的实证评估。这项工作预期的主要结果是改进了对基于程序分析的测试和维护的理解和改进的技术--基于可靠的经验数据。这项工作还将提供一个公开可用的基础设施，它将支持未来在整个软件工程社区中进行基于程序分析的测试和维护技术的工具开发、试验和教育。

Maintenance and testing of software are expensive. Program analysis techniques, which analyze source code to obtain information about software, promise to reduce this expense by providing information that helps software engineers perform specific maintenance and testing tasks such as program understanding, impact analysis, and regression testing. However, to be adopted or even considered for adoption, program-analysis-based testing and maintenance techniques must be empirically demonstrated to scale to large software systems. The primary goal of this reqearch is to empirically investigate the applicability of program- analysis-based testing and maintenance techniques to large- scale systems. The research involves three overall efforts: (1) construction of an extensible experimental infrastructure including a software system and repository of subject systems; (2) development of program-analysis techniques that support testing and maintenance of large- scale systems; (3) rigorous empirical evaluation of the infrastructure, and of the program-analysis-based testing and maintenance techniques. The principal results expected from this work are an improved understanding of, and improved techniques for, program-analysis-based testing and maintenance -- founded on hard empirical data. The work will also provide a publicly available infrastructure, that will support future tool development for, experimentation with, and education in program-analysis- based testing and maintenance techniques throughout the software engineering community.