Parallelism in Language Translation

这是对可以在多大程度上使用并行性来加快某些软件开发任务的执行的调查。在软件开发环境中，以多处理器工作站的形式或以(单处理器)工作站的网络形式的并行性正变得普遍可用。这项研究试图利用并行性的可用性来加速现有的软件任务，并考虑使用更雄心勃勃的软件工具。重点是执行语言翻译的软件工具，包括优化的传统编程语言翻译，以及文本格式化、校对和可被视为实现上下文无关语言翻译的各种其他软件工具。这些语言翻译问题被指定为属性文法。它们提供了翻译过程的高级功能规范，允许自动构建高效的并行翻译器。主要有三个问题：1.高效并行属性文法的计算策略及其相对于顺序计算器和更传统的编译技术的性能。2.在属性文法框架中规范各种全局编译器优化技术。3.对属性文法用于说明与传统程序设计语言翻译无关的各种问题的通用性进行评估，包括将其扩展用于图而不是树上的计算说明。

This is an investigation of the extent to which parallelism can be used to speed up the execution of certain software development tasks. Parallelism is becoming commonly available in software development environments, in the form of multiprocessor workstations or in the form of networks of (uniprocessor) workstations. This research attempts to capitalize on the availability of parallelism to speed up existing software tasks as well as to contemplate the use of more ambitious software tools. Concentration is on software tools for performing language translation, including traditional programming language translation with optimization, as well as text formatting, proof checking and a variety of other software tools that can be viewed as implementing the translation of a context-free language. These language translation problems are specified as attribute grammars. These provide a high-level, functional specification of the translation process which allows the automatic construction of efficient parallel translators. There are three principal issues: 1. Strategies for efficient parallel attribute grammar evaluation and their performance relative to sequential evaluators and more traditional compilation techniques. 2. Specification of various global compiler optimization techniques in an attribute grammar framework. 3. An assessment of the generality of attribute grammars for specification of various problems not related to traditional programming language translation, including extending their use for specifications of computations on graphs rather than on trees.