Generator of efficient strongly typed abstract syntax trees in Java

Abstract

C e n t r u m v o o r W i s k u n d e e n I n f o r m a t i c a Software ENgineeringA generator of efficient strongly typed abstract syntax trees in Java A generator of efficient strongly typed abstract syntax trees in Java ABSTRACT Abstract syntax trees are a very common data-structure in language related tools. For example compilers, interpreters, documentation generators, and syntax-directed editors use them extensively to extract, transform, store and produce information that is key to their functionality. We present a Java back-end for ApiGen, a tool that generates implementations of abstract syntax trees. The generated code is characterized by strong typing combined with a generic interface and maximal sub-term sharing for memory efficiency and fast equality checking. The goal of this tool is to obtain safe and efficient programming interfaces for abstract syntax trees. The contribution of this work is the combination of generating a strongly typed data-structure with maximal sub-term sharing in Java. Practical experience shows that this approach can not only be used for tools that are otherwise manually constructed, but also for internal datastructures in generated tools. Abstract. Abstract syntax trees are a very common data-structure in language related tools. For example compilers, interpreters, documentation generators, and syntax-directed editors use them extensively to extract, transform, store and produce information that is key to their functionality. We present a Java back-end for ApiGen, a tool that generates implementations of abstract syntax trees. The generated code is characterized by strong typing combined with a generic interface and maximal sub-term sharing for memory efficiency and fast equality checking. The goal of this tool is to obtain safe and efficient programming interfaces for abstract syntax trees. The contribution of this work is the combination of generating a strongly typed data-structure with maximal sub-term sharing in Java. Practical experience shows that this approach can not only be used for tools that are otherwise manually constructed, but also for internal data-structures in generated tools. ACM Computing Classification

Performance Modeling of Maximal Sharing

“…is a prime source of inspiration [22,24,25]. Both in C and in Java this is a successful library that employs maximal sharing for representing atomic data-types, lists and trees.…”

“…Potential adopters of maximal sharing suffer from a common problem: converting a library to use maximal sharing is hard and costly [22,24,25]: it is a cross-cutting design decision with difficult to tune implementation details. To illustrate one of many pitfalls, let us take a Java library for graph processing as example.…”

A Generic Framework for Automated Quality Assurance of Software Models –Implementation of an Abstract Syntax Tree

“…However, the usage of ASTs now encompasses the implementation of many language related tools, such as interpreters, document generators and syntax-directed editors, etc. [8].…”

“…One such use that an AST can support is in duplicate code detection, whereby an AST designed to support data matching efficiency [8] requires only a pattern to be found. The significance of this is that only an initial node needs to be identified which is subsequently followed by a predetermined pattern of nodes.…”

Tom: Piggybacking Rewriting on Java

“…Tom is the result of a long term effort, on one hand integrating innovative ideas and concepts, and on the other hand combining and incorporating key notions and techniques developed in well reputed research teams. Regarding the data-structures, the implementation of the generator has been done in strong cooperation with the authors of ApiGen [15] and followed our experience with the ATerm [14] library used by ASF+SDF. The originality of our solution is to provide typed constructs resulting in a faster and safer implementation.…”