Dynamically Extensible Computational Reflection

点击翻译按钮获取中文摘要

Reflection has been explored in an interpreted framework and has tended to assume the availability of particular pieces of the state of a program's interpretation, even the current source code expression. Instead of programs, this research emphasizes computations as the primary elements of reflection. The research has two facets. The first is to formalize and continue the theoretical development of a system for computational reflection that is free from interpreter-based constraints. This system supports the introduction of multiple user-defined reflective properties that can be designed independently and combined in arbitrary fashion. Furthermore, new reflective properties can be introduced after the programs in which they are to be used have already been compiled. The second facet is to explore the many uses of such a system. In addition to the simple, but powerful, uses already explored, the system is expected to be useful for rapid prototyping of language features, collaborative software development, and static analysis of programs for intense optimization. Prior research in programming with first-class contexts has investigated the meaning and use of static meta-level contextual information. The word static here is intended to refer to that information available in the text of a program, without running it. The current research extends the previous work by exploring the meaning and use of dynamic meta-level contextual information, which is that information available in the meta-level of a running process. As in the previous work, this project emphasizes modularity, allowing many different kinds of meta-level knowledge to be utilized simultaneously. ***