程序覆盖进行错误自动定位方法的在线实验平台  

Online Experiment Platform for Coverage-based Fault Localization Techniques

作　　者：刘誉臻 孙文辉[2] 张震宇[1] 石琼[3] 

机构地区：[1]中国科学院软件研究所,北京100080 [2]北京交通大学计算机与信息技术学院,北京100044 [3]中北大学电子与计算机科学技术学院,太原030051

出　　处：《小型微型计算机系统》2016年第11期2397-2401,共5页Journal of Chinese Computer Systems

基　　金：国家自然科学基金项目(61379045)资助

摘　　要：软件错误定位是软件测试领域的研究热点.基于覆盖的软件错误自动定位技术,利用概率模型对程序执行信息进行统计分析,估计程序语句与错误的相关度,预测错误位置.其最基本的思想是,错误语句往往较多地被失败用例覆盖而较少地被成功用例覆盖.此类技术层出不穷,演化迅速,往往需要试验各种公式,并辅以繁杂的测试分析.由于研究人员的差异和实验条件的多样,实验中往往需要反复进行测试和重新编写脚本;工具和经验共享不利,已成为阻碍软件错误定位研究发展的不利条件.本文设计实现了针对软件错误定位技术的在线评测和实验平台,能简化实验步骤、避免编写脚本、设计公式原型、调整实验参数并自动生成结果.目前已有多家科研单位利用本文实现的平台进行着此类研究.Automated software fault localization is a hot topic in software testing. Coverage-based fault localization techniques evaluate the suspiciousness of a statement being related to fault by investigating its coverage information collected in program executions. The basic idea is that faulty statements are mostly covered by failed program runs, and seldom covered by passed program runs. Such tech- niques appear and evolve very fast,which design and implementation consists of tedious formula tuning and complicated empirical studies including repeatedly nmning test cases, setting up different experiments, re-composing test scripts. Because of the low efficiency of experience sharing and reusing of test platforms, the development of the study has been seriously affected. In this work, we introduce our online-judge platform for coverage-based software fault localization techniques. Our platform simplifies the experiment setup step, avoids the composing of scripts,provides UIs to tune formulas and adjust parameters,and automatically generates intuitive results. Our platform has been used to conduct fault localization studies by some universities.

关 键 词：软件测试 错误定位 程序覆盖 公式编辑 实验平台 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]