The Spineless Tagless G-machine is an abstract machine designed to support nonstrict higher-order functional languages. This presentation of the machine falls into three parts. Firstly, we give a general discussion of the design issues involved in implementing non-strict functional languages. Next, we present the STG language, an austere but recognisably-functional language, which as well as a denotational meaning has a well-defined operational semantics. The STG language is the "abstract machine code" for the Spineless Tagless G-machine. Lastly, we discuss the mapping of the STG language onto stock hardware. The success of an abstract machine model depends largely on how efficient this mapping can be made, though this topic is often relegated to a short section. Instead, we give a detailed discussion of the design issues and the choices we have made. Our principal target is the C language, treating the C compiler as a portable assembler. Version 2.5 of this paper (minus appendix) appe...

We give an overview of the Glasgow Haskell compiler, focusing especially on way in which we have been able to exploit the rich theory of functional languages to give very practical improvements in the compiler. The compiler is portable, modular, generates good code, and is freely available. 1 Introduction Computer Science is both a scientific and an engineering discipline. As a scientific discipline, it seeks to establish generic principles and theories that can be used to explain or underpin a variety of particular applications. As an engineering discipline, it constructs substantial artefacts of software and hardware, sees where they fail and where they work, and develops new theory to underpin areas that are inadequately supported. (Milner [1991] eloquently argues for this dual approach in Computer Science. ) Functional programming is a research area that offers an unusually close interplay between these two aspects (Peyton Jones [1992b]). Theory often has immediate practical appl...

Abstract not found

This paper investigates the transformation of v -terms into continuation-passing style (CPS). We show that by appropriate j-expansion of Fischer and Plotkin's two-pass equational specification of the CPS transform, we can obtain a static and context-free separation of the result terms into "essential" and "administrative" constructs. Interpreting the former as syntax builders and the latter as directly executable code, we obtain a simple and efficient one-pass transformation algorithm, easily extended to conditional expressions, recursive definitions, and similar constructs. This new transformation algorithm leads to a simpler proof of Plotkin's simulation and indifference results. Further we show how CPS-based control operators similar to but more general than Scheme's call/cc can be naturally accommodated by the new transformation algorithm. To demonstrate the expressive power of these operators, we use them to present an equivalent but even more concise formulation of t...

Many compilers do some of their work by means of correctness-preserving, and hopefully performance-improving, program transformations. The Glasgow Haskell Compiler (GHC) takes this idea of "compilation by transformation" as its war-cry, trying to express as much as possible of the compilation process in the form of program transformations. This paper reports on our practical experience of the transformational approach to compilation, in the context of a substantial compiler. The paper appears in the Proceedings of the European Symposium on Programming, Linkoping, April 1996. 1 Introduction Using correctness-preserving transformations as a compiler optimisation is a well-established technique (Aho, Sethi & Ullman [1986]; Bacon, Graham & Sharp [1994]). In the functional programming area especially, the idea of compilation by transformation has received quite a bit of attention (Appel [1992]; Fradet & Metayer [1991]; Kelsey [1989]; Kelsey & Hudak [1989]; Kranz [1988]; Steele [1978]). A ...

Virtually every compiler performs transformations on the program it is compiling in an attempt to improve efficiency. Despite their importance, however, there have been few systematic attempts to categorise such transformations and measure their impact. In this paper we describe a particular group of transformations --- the "let-floating" transformations --- and give detailed measurements of their effect in an optimising compiler for the non-strict functional language Haskell. Let-floating has not received much explicit attention in the past, but our measurements show that it is an important group of transformations (at least for lazy languages), offering a reduction of more than 30% in heap allocation and 15% in execution time.

In this thesis we present and analyse a set of automatic source-to-source program transformations that are suitable for incorporation in optimising compilers for lazy functional languages. These transformations improve the quality of code in many different respects, such as execution time and memory usage. The transformations presented are divided in two sets: global transformations, which are performed once (or sometimes twice) during the compilation process; and a set of local transformations, which are performed before and after each of the global transformations, so that they can simplify the code before applying the global transformations and also take advantage of them afterwards. Many of the local transformations are simple, well known, and do not have major effects on their own. They become important as they interact with each other and with global transformations, sometimes in non-obvious ways. We present how and why they improve the code, and perform extensive experiments wit...

syntax of the source language ` c : ' f:::; x : ø ; :::g ` x : ø ß ` e : ø !ø ß ` fix e : ø ß [ fx : ø 1 g ` e : ø 2 ß ` x : ø 1 : e : ø 1 !ø 2 ß ` e 0 : ø 1 !ø 2 ß ` e 1 : ø 1 ß ` @ e 0 e 1 : ø 2 ß ` e 1 : ' ß ` e 2 : ø ß ` e 3 : ø ß ` if e 1 then e 2 else e 3 : ø ß ` e 0 : ø 0 ß [ fx : ø 0 g ` e 1 : ø 1 ß ` let x = e 0 in e 1 : ø 1 ß ` e 1 : ø 1 ß ` e 2 : ø 2 ß ` pair e 1 e 2 : ø 1 \Theta ø 2 ß ` e : ø 1 \Theta ø 2 ß ` fst e : ø 1 ß ` e : ø 1 \Theta ø 2 ß ` snd e : ø 2 Fig. 2. Type-checking rules for the source language approach is used by Kesley and Hudak [11] and by Fradet and Le M'etayer [9]. Both include a CPS transformation. Fradet and Le M'etayer compile both CBN and CBV programs by using the CBN and the CBV CPS-transformation. Recently, Burn and Le M'etayer have combined this technique with a global programanalysis [2], which is comparable to our goal here. 1.4 Overview Section 2 presents the syntax of the source language and the strictness-annotated language. We c...

Stratego is a domain-specic language for the specication of program transformation systems. The design of Stratego is based on the paradigm of rewriting strategies: user-denable programs in a little language of strategy operators determine where and in what order transformation rules are (automatically) applied to a program. The separation of rules and strategies supports modularity of specications. Stratego also provides generic features for specication of program traversals. In this paper we present a case study of Stratego as applied to a non-trivial problem in program transformation. We demonstrate the use of Stratego in eliminating intermediate data structures from (also known as deforesting) functional programs via the warm fusion algorithm of Launchbury and Sheard. This algorithm has been specied in Stratego and embedded in a fully automatic transformation system for kernel Haskell. The entire system consists of about 2600 lines of specication code, which bre...

We present a method, adapted to polymorphically typed functional languages, to detect and collect more garbage than existing GCs. It can be applied to strict or lazy higher order languages and to several garbage collection schemes. Our GC exploits the information on utility of arguments provided by polymorphic types of functions. It is able to detect garbage that is still referenced from the stack and may collect useless parts of otherwise useful data structures. We show how to partially collect shared data structures and to extend the type system to infer more precise information. We also present how this technique can plug several common forms of space leaks.