SHF: Small: Dynamic Analysis on Code Fragments

在许多软件工程环境中，软件开发人员希望理解、测试、调试和验证相对较小的代码片段，而不是整个程序。然而，目前可用的程序分析和测试工具仅适用于完整的程序，因为它们可以作为一个完整的程序进行编译和执行。当目标是仅以小的代码片段为目标时，在整个程序的上下文中测试和分析目标代码是过于昂贵和耗时的，这需要配置整个系统、设置执行环境并找到提供执行整个程序所需的所有数据值的测试输入套件。如果可以将感兴趣的代码片段的程序属性作为独立的代码单元进行检查，则软件开发过程将更加高效和有效。这项研究项目解决了将代码片段转换为可测试单元的技术挑战。如果软件测试和分析能够成功地应用于代码片段，而不是整个程序，那么开发正确代码的整个过程将得到简化。本项目侧重于动态程序分析，这是通过在运行时环境中执行程序来执行的计算机软件分析(与静态分析相反，静态分析是对源代码的分析)。该项目将生成能够促进程序分析工具和软件工程实践的最先进水平的算法、工具和数据。结果将通过会议、课堂、开源项目、行业合作和STEM志愿者机会传播。为了使动态程序分析有效，目标程序必须在执行时具有足够的测试输入，以产生有趣的(例如，不正确或异常的)行为。由于经常有大量的程序路径和必须测试的大量输入，使得动态分析变得困难。通常，测试是无效的，因为它们产生了压倒性的假阳性，并且经常在程序的某些部分失败，而这些部分甚至与理解、调试或验证目标代码的目的无关。该项目将开发一种方法，以获取从原始程序中选择或构造的代码片段，并使用新的语法修补技术生成可编译和可执行单元，以实现对代码片段的动态分析。它还将开发一种技术来搜索和选择要操作的有意义的代码片段。该奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

In many software engineering environments, software developers would like to understand, test, debug and verify a relatively small fragment of code instead of the entire program. However, currently available program analysis and testing tools only work on programs that are whole, in the sense that they can be compiled and executed as a whole program. When the goal is to target only a small code fragment, it is excessively expensive and time-consuming to test and analyze the targeted code in the context of entire programs, which requires configuring a whole system, setting up the execution environment, and finding suites of test inputs that supply all the needed data values to execute the whole program. Software development processes would be much more efficient and effective if a program property for the code fragment of interest could be checked as a standalone unit of code. This research project addresses the technical challenges in converting code fragments into testable units. If software testing and analysis could be applied successfully to code fragments, rather than whole programs, the overall process of developing correct code would be streamlined. This project focuses on dynamic program analysis, which is the analysis of computer software that is performed by executing programs in a run-time environment (as opposed to static analysis, which is analysis of the source code). The project will generate algorithms, tools, and data that can advance the state of the art of program analysis tools and software engineering practice. The results will be disseminated through conferences, classrooms, open source projects, industrial collaborations and STEM volunteer opportunities.For dynamic program analysis to be effective, the target program must be executed with sufficient test inputs to produce interesting (e.g., incorrect or anomalous) behaviors. Dynamic analysis is made difficult by frequently huge numbers of program paths, and the large number of inputs that must be tested. Often, the tests are ineffective because they produce an overwhelming number of false positives and often fail on parts of the program that are not even relevant to the purpose of understanding, debugging or verifying the targeted code. The project will develop an approach to take code fragments, selected or constructed from the original program, and generate compilable and executable units, using new syntactic patching techniques to enable dynamic analysis on code fragments. It will also develop a techniques to search for and select meaningful code fragments on which to operate.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.