Framework, Algorithms, and Applications for Cross-Module Inlining

现代编程语言鼓励程序员使用抽象和封装来向其他模块隐藏数据表示细节。但这会在没有内联扩展的情况下导致执行速度较慢(用过程体替换过程调用)。这个问题在支持方法和函数抽象的面向对象和函数式编程语言中尤为严重。大多数以前关于内联扩展的工作(在命令式语言中)都无法内联高阶和递归函数，但在函数式语言(以及具有动态方法的面向对象语言)中，这些可能是至关重要的。由于模块可以彼此分开编译，跨模块的内联扩展使程序链接复杂化。这项工作使用lambda演算框架进行跨模块内联扩展，它将编译管理问题与内联算法完全分开。本项目的目标是实现该框架，然后在该框架内设计可在现场测试的内联算法。*

Modern programming languages encourage programmers to use abstraction and encapsulation to hide one moduleOs data representation details from other modules. But this leads to slow execution without in-line expansion (substitution of procedure bodies for procedure calls). The problem is particularly acute in object-oriented and functional programming languages that support method and function abstraction. Most previous work on in-line expansion (in imperative languages) fails to inline higher-order and recursive functions, but in functional languages (and object- oriented languages, with dynamic methods) these can be crucial. Since modules can be compiled separately from each other, cross-module in-line expansion complicates program linkage. This work uses a lambda-calculus framework for cross-module in-line expansion, which cleanly separates the compilation- management issues from the inlining algorithms. The objective of this project is to implement this framework, and then design inlining algorithms within the framework that can be tested in the field.***