EAGER: Investigating the Feasibility and Scope of Automated Refactoring for Distributed Software

随着软件在其生命周期内的维护和新功能的添加，经常需要重新构建软件以继续使用预期的功能和性能。这种重组称为重构，是乏味的，容易出错，消耗软件开发时间并延迟发布。在现代软件开发过程中，重构是常规的，但只适用于在一台机器上运行的软件。随着分布式应用程序的日益普及，例如用于Web、移动的系统和物联网的应用程序，重构技术不再适用。这个EAGER基金会探索了将重构扩展到分布式系统所涉及的问题。重构是应用保持语义的程序转换来重构软件，以获得更好的性能、模块化和其他属性。重构已成为现代软件开发过程中不可或缺的一部分，但仅适用于在单个地址空间内运行的集中式软件。当开发分布式系统时，程序员将现有的重构技术应用于其各个组件，将它们视为独立的、不相关的部分，因此这些技术仍然不知道这些部分之间的分布式通信。当程序员只能重构分布式系统的各个部分时，（1）整个系统的执行语义可能无法保留，（2）系统的分布无法改变，因此无法实现修改现代分布式系统的许多重要目标。分布式应用程序有许多随之而来的挑战-例如由独立程序员在分散的生态系统中编写，以及使用具有不同语义的多种语言编写，这些语言需要相关和保留。EAGER奖将揭示技术可能性和挑战，作为理解和应对这些挑战的一步。该奖项反映了NSF的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

As software is maintained over its lifetime and new features are added, it often becomes necessary to restructure the software to continue working with expected functionality and performance. This restructuring, called refactoring, is tedious and error-prone, consuming software development time and delaying releases. In modern software development processes, refactoring is routine but only applies to software running on one machine. With the increasing ubiquity of distributed applications, such as those for the web, mobile systems, and the Internet of Things, refactoring techniques do not apply. This EAGER grant explores the issues involved in extending refactoring to distributed systems. Refactoring is the application of semantics-preserving program transformations to restructure software for better performance, modularity, and other attributes. Refactoring has become an integral part of the modern software development processes, but only for centralized software that runs within a single address space. When evolving a distributed system, programmers apply existing refactoring techniques to its individual components, treating them as separate, unrelated parts, so these techniques remain unaware of the distributed communication across these parts. When programmers can refactor only the individual parts of a distributed system, (1) the overall system’s execution semantics may not be preserved, and (2) the system’s distribution cannot be changed, thus making it impossible to fulfill many important objectives for modifying modern distributed systems. Distributed applications have many accompanying challenges – such as being written by independent programmers in decentralized ecosystems and in multiple languages with different semantics that need to be related and preserved. This EAGER grant will uncover technical possibilities and challenges as a step toward understanding and addressing these challenges.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.