Programming with First-Class Contexts

9302114弗里德曼上下文是表达跨多个词汇作用域的变量共享的完美方法。上下文是一个语法树，它具有称为空洞的非末尾叶子，稍后将用术语(这些术语也是上下文)填充。与替换不同，用术语填充上下文的空洞会导致术语的自由变量被上下文的绑定变量捕获。因此，一流的上下文是信息共享程序片段增量开发的基础。赋予上下文一流的地位便于单独编译上下文。目标代码可以在上下文的自由变量的不同实例化上重用，或者在解析其不同填充术语的自由变量时重用。上下文编程将编译、链接(上下文填充)和评估置于用户的控制之下。它是一个概念简单但非常通用的编程范例。特别是，模块化编程和面向对象编程都是上下文编程的特例。本研究包括正式研究和实验研究两部分。形式推理系统提供了一个分析和比较上下文编程与其他编程语言概念和机制的平台。实验有望产生有洞察力的信息，以评估设计决策，发现缺陷，并发展更好的理解上下文编程。***

9302114 Friedman Contexts are a perfect means for expressing variable sharing across multiple lexical scopes. A context is a syntax tree with nonterminal leaves, called holes, that are to be filled later with terms (which are contexts as well). Unlike substitution, filling a context's hole with a term causes the term's free variables to be captured by the context's bound variables. First- class contexts thus are a basis for incremental development of information-sharing program pieces. Giving contexts first-class status facilitates separate compilation of contexts. The target code can be reused on different instantiations of the context's free variables or be reused on resolving its different filling terms' free variables. Contextual programming brings compilation, linking (context filling), and evaluation under the user's control. It is a conceptually simple and yet very general programming paradigm. In particular, modular and object-oriented programming are both special cases of contextual programming. This research has both formal and experimental components. Formal reasoning systems furnish a platform for analyzing and comparing contextual programming with other programming language concepts and mechanisms. Experiments are expected to generate insightful information to evaluate design decisions, uncover deficiencies, and develop a better understanding of contextual programming. ***