DMCA
A type-correct, stack-safe, provably correct, expression compiler (2006)
Cached
Download Links
- [www.dcs.st-and.ac.uk]
- [www.eecs.harvard.edu]
- CiteULike
Other Repositories/Bibliography
| Venue: | in Epigram. Submitted to the Journal of Functional Programming |
| Citations: | 19 - 0 self |
BibTeX
@INPROCEEDINGS{Mckinna06atype-correct,,
author = {James Mckinna and Joel Wright},
title = {A type-correct, stack-safe, provably correct, expression compiler},
booktitle = {in Epigram. Submitted to the Journal of Functional Programming},
year = {2006}
}
Share
 |
 |
 |
 |
||
OpenURL
Abstract
Conventional approaches to compiler correctness, type safety and type preservation have focused on off-line proofs, either on paper or formalised with a machine, of existing compilation schemes with respect to a reference operational semantics. This pearl shows how the use of dependent types in programming, illustrated here in Epigram, allows us not only to build-in these properties, but to write programs which guarantee them by design and subsequent construction. We focus here on a very simple expression language, compiled into tree-structured code for a simple stack machine. Our purpose is not to claim any sophistication in the source language being modelled, but to show off the metalanguage as a tool for writing programs for which the type preservation and progress theorems are self-evident by construction, and finally, whose correctness can be proved directly in the system. In this simple setting we achieve the following; • a type-preserving evaluation semantics, which takes typed expressions to typed values. • a compiler, which takes typed expressions to stack-safe intermediate code. • an interpreter for compiled code, which takes stack-safe intermediate code to a big-step stack transition. • a compiler correctness proof, described via a function whose type expresses the equational correctness property. 1
Keyphrases
expression compiler stack-safe intermediate code type preservation typed expression simple setting type safety dependent type conventional approach compiled code off-line proof tree-structured code subsequent construction simple stack machine equational correctness property simple expression language source language big-step stack transition reference operational semantics compilation scheme type-preserving evaluation semantics compiler correctness proof progress theorem
