Towards Practical, Expressive Multimethods

走向实用、富有表现力的多种方法

• 批准号: 9970986
• 负责人: Craig Chambers
• 金额: $ 20万
• 依托单位: University of Washington
• 依托单位国家: 美国
• 项目类别: Continuing Grant
• 财政年份: 1999
• 资助国家: 美国
• 起止时间: 1999-09-01 至 2002-08-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=9970986&HistoricalAwards=false
• 关键词: Towards; Practical; Expressive; Multimethods

9970986 Chambers, Craig D. University of Washington, SeattleTowards Practical, Expressive MultimethodsOne of the promises of object-oriented programming is the development of highly reusable and extensible software components, but traditional object-oriented languages have fundamental limitations that hinder the attainment of this goal. Multimethods, as found in a few object-oriented languages, offer several advantages in uniformity, flexibility, reusability, and extensibility. However, multimethod in current languages suffer from several weaknesses compared to traditional object-oriented languages, including a lack of modular approaches to static typechecking, a lack of strong encapsulation mechanisms, and poorer run-time performance for message dispatching. This project seeks to solve these problems. A flexible model fordispatching is being developed that subsumes single dispatching, multiple dispatching, and functions defined by pattern-matching, along with a classless object model and a static type system. A module system is being added to this base, to provide encapsulation and to define the units of separate static typechecking. New implementation techniques for efficient message dispatching are also being developed. To assess this research, the design is being embodied into both a small core language suitable for theoretical study (named Dubious) and an extended language suitable for practical programming (named Diesel), and the Vortex optimizing compiler infrastructure is being modified to include the new dispatching algorithms.

9970986钱伯斯，克雷格·D·华盛顿大学，西雅图面向对象编程的前景之一是开发高度可重用和可扩展的软件组件，但传统的面向对象语言有根本的局限性，阻碍了这一目标的实现。正如在一些面向对象语言中发现的那样，多方法在统一性、灵活性、可重用性和可扩展性方面提供了几个优势。然而，与传统的面向对象语言相比，当前语言中的多方法存在几个弱点，包括缺乏用于静态类型检查的模块化方法，缺乏强大的封装机制，以及消息分派的运行时性能较差。本项目旨在解决这些问题。正在开发一种灵活的调度模型，该模型包括单调度、多调度和由模式匹配定义的功能，以及无类对象模型和静态类型系统。一个模块系统正被添加到这个库中，以提供封装并定义单独的静态类型检查的单元。还在开发用于高效消息分派的新实现技术。为了评估这项研究，设计既体现为适合理论研究的小型核心语言(称为可疑语言)，也体现为适合实际编程的扩展语言(称为柴油)，并且正在修改Vortex优化编译器基础设施，以包括新的调度算法。