Analysis of MiniJava Programs via Translation to ML

TL;DR

This paper proposes translating MiniJava programs into ML to leverage ML's analysis tools for verifying Java-like programs, aiming to improve reasoning techniques and tool robustness across language styles.

Contribution

It introduces a translation method from MiniJava to ML that uses only core ML features, enabling cross-language program analysis and verification.

Findings

Developed a translation from MiniJava to ML

Demonstrated potential for using ML tools for Java-like program analysis

Outlined future work for improving analysis techniques

Abstract

MiniJava is a subset of the object-oriented programming language Java. Standard ML is the canonical representative of the ML family of functional programming languages, which includes F# and OCaml. Different program analysis and verification tools and techniques have been developed for both Java-like and ML-like languages. Naturally, the tools developed for a particular language emphasise accurate treatment of language features commonly used in that language. In Java, this means objects with mutable properties and dynamic method dispatch. In ML, this means higher order functions and algebraic datatypes with pattern matching. We propose to translate programs from one language into the other and use the target language's tools for analysis and verification. By doing so, we hope to identify areas for improvement in the target language's tools and suggest techniques, perhaps as used in the source language's tools, that may guide their improvement. More generally, we hope to develop tools for reasoning about programs that are more resilient to changes in the style of code and representation of data. We begin our programme by outlining a translation from MiniJava to ML that uses only the core features of ML; in particular, it avoids the use of ML's mutable references.