CAREER: Differential Types and Declarative Hypothesis Testing for Software Evolution

软件在第一个版本发布后很长时间内仍在继续发展。在演化过程中对源代码所做的更改可能不具有预期的语义（修复错误，添加功能），这对软件生产者和软件消费者都产生了负面影响，并且具有很高的经济成本。更一般地说，软件演化过程还没有得到很好的理解。 为了纠正这种情况，这项工作将使用来自两种方式分析软件历史的信息。首先，开发一种基于类型系统的方法，称为差分类型，它可以在更改之前和之后对源代码的属性进行推理，并在开发时揭示源代码更改的语义。 第二，构建一个声明性假设检验的框架，自动化和简化软件演化和软件工程的经验研究过程，这项工作的跨学科基础将促进两个主要社区，编程语言和软件工程之间的互动。 这项研究的结果将导致更好地了解软件如何演变，并在软件演化过程中的改进。通过在部署软件之前使软件改变的语义明确并且揭示某些种类的错误，减少了软件错误和不正确的软件更新的发生率以及与之相关联的成本。该项目将改变研究人员进行实证软件进化研究的方式，将重点从如何测试假设转移到正确的假设是什么。 这些贡献将被公开，这不仅有利于研究社区，也有利于软件开发社区。这项工作的想法将被纳入教育，以便明天的毕业生更好地准备应对软件进化的挑战。

Software continues to evolve long after the first version is released. Changes made to source code in the course of evolution might not have the intended semantics (fix bugs, add features), which negatively impacts both software producers and software consumers, and has high economic costs. More generally, the software evolution process is not very well understood. To remedy this situation, this work will use information derived from analyzing software history in two ways. First, to develop a type systems-based approach named differential typing, that enables reasoning about the properties of source code before and after a change has been made, and reveals the semantics of source code changes at development time. Second, to construct a framework for declarative hypothesis testing that automates and simplifies the process of conducting empirical studies in software evolution, and software engineering in general.The interdisciplinary foundation of this work will catalyze interaction between two primary communities, programming languages and software engineering. The results of this research will lead to a better understanding of how software evolves, and to improvements in the software evolution process. By making the semantics of software changes explicit and revealing certain kinds of errors before software is deployed, the incidence of, and costs associated with, software bugs and incorrect software updates are reduced. The project will transform the way researchers conduct empirical software evolution studies by shifting the focus from how hypotheses are tested to what the right hypotheses are. The contributions will be made publicly available, which will benefit not only the research community, but also the software development community. Ideas from this work will be incorporated into education, so that tomorrow's graduates are better prepared to tackle software evolution challenges.