Program transformation in meta programming environment

点击翻译按钮获取中文摘要

We are developing a system called meta computing environment, in which several computation models coexist sharing common resources including stacks and heap. Presently MC comprises two models, namely reduction and resolution based models. LISP and Prolog are constructed on each models. We observe program transformations play an important role in reasoning about programs, especially so when we are interested in meta-theorems of programs.Our results are summarized in the following points. (i) Development of program transformation schemes in functional programs and application of the schemes to compilation of LISP programs. LISP compiler successively transform programs; first LISP-to-LISP, then LISP to categorical expressions and finally code sequences of an abstract machine. The process of the transformations are formally described. (ii) Development of type inference system for LISP programs. Prototype implementation of a type inference system is constructed; it serves as a front end processor which generates typed LISP programs. (iii) Methodologies for meta programming. Since meta computing environment has LISP and Prolog, we used features of the both in performing researches outlined in (i) and (ii). Pattern matching and backtracking are taken advantage of in program transformation on one hand, and on the other hand programs with more refined control structures are expressed in LISP.