Software engineers are faced with a dilemma. They want to write general and wellstructured programs that are flexible and easy to maintain. On the other hand, generality has a price: efficiency. A specialized program solving a particular problem is often significantly faster than a general program. However, the development of specialized software is time-consuming, and is likely to exceed the production of today’s programmers. New techniques are required to solve this so-called software crisis. Partial evaluation is a program specialization technique that reconciles the benefits of generality with efficiency. This thesis presents an automatic partial evaluator for the Ansi C programming language. The content of this thesis is analysis and transformation of C programs. We develop several analyses that support the transformation of a program into its generating extension. A generating extension is a program that produces specialized programs when executed on parts of the input. The thesis contains the following main results.

We describe a design pattern for writing programs that traverse data structures built from rich mutually-recursive data types. Such programs often have a great deal of "boilerplate" code that simply walks the structure, hiding a small amount of "real" code that constitutes the reason for the traversal. Our technique allows most...

In functional programming, intermediate data structures are often used to "glue" together small programs. Deforestation is a program transformation to remove these intermediate data structures automatically. We present a simple algorithm for deforestation based on two fusion rules for hylomorphism, an expressive recursion pattern. A generic notation for hylomorphisms is introduced, where natural transformations are explicitly factored out, and it is used to represent programs. Our method successfully eliminates intermediate data structures of any algebraic type from a much larger class of compositional functional programs than previous techniques. 1 Introduction In functional programming, programs are often constructed by "gluing" together small components, using intermediate data structures to convey information between them. Such data are constructed in one component and later consumed in another component, but never appear in the result of the whole program. The compositional styl...

Note: This document accompanies the paper “Practical type inference for arbitrary-rank types ” [6]. Prior reading of the main paper is required. 1 Contents

A number of useful optimisations are enabled if we can determine when a value is accessed at most once. We extend the Hindley-Milner type system with uses, yielding a typeinference based program analysis which determines when values are accessed at most once. Our analysis can handle higher-order functions and data structures, and admits principal types for terms. Unlike previous analyses, we prove our analysis sound with respect to call-by-need reduction. Call-by-name reduction does not provide an accurate model of how often a value is used during lazy evaluation, since it duplicates work which would actually be shared in a real implementation. Our type system can easily be modified to analyse usage in a call-by-value language. 1 Introduction This paper describes a method for determining when a value is used at most once. Our method is based on a simple modification of the Hindley-Milner type system. Each type is labelled to indicate whether the corresponding value is used at most onc...

Program fusion is the process whereby separate pieces of code are fused into a single piece, typically transforming a multi-pass algorithm into a single pass. Recent work has made it clear that the process is especially successful if the loops or recursions are expressed using catamorphisms (e.g.foldr) and constructor-abstraction (e.g. build). In this paper we show how to transform recursive programs into this form automatically, thus enabling the fusion transformation to be applied more easily than before. 1 Introduction There are significant advantages to multi-pass algorithms, in which intermediate data-structures are created and traversed. In particular, each of the passes may be relatively simple, so are both easier to write and are potentially more reusable. By separating many distinct phases it becomes possible to focus on a single task, rather than attempting to do many things at the same time. The classic toy example of this is to compute the sum of the squares of the numbe...

We describe the design and current status of our effort to implement the programming model of nested data parallelism into the Glasgow Haskell Compiler. We extended the original programmingmodel and its implementation, both of which were first popularised by the NESL language, in terms of expressiveness as well as efficiency. Our current aim is to provide a convenient programming environment for SMP parallelism, and especially multicore architectures. Preliminary benchmarks show that we are, at least for some programs, able to achieve good absolute performance and excellent speedups.

Synchronous data-flow is a programming paradigm which has been successfully applied in reactive systems. In this context, it can be characterized as some class of static bounded memory data-flow networks. In particular, these networks are not recursively defined, and obey some kind of "synchronous" constraints (clock calculus). Based on Kahn's relationship between data-flow and stream functions, the synchronous constraints can be related to Wadler's listlessness, and can be seen as sufficient conditions ensuring listless evaluation. As a by-product, those networks enjoy efficient compiling techniques. In this paper, we show that it is possible to extend the class of static synchronous data-flow to higher order and dynamical networks, thus giving sense to a larger class of synchronous data-flow networks.

this paper, we propose an algorithm which can automatically turn all practical recursive definitions into structural hylomorphisms making program fusion be easily applied. 1 Introduction

Languages like ML and Haskell encourage the view of values as first-class entities that can be passed as arguments or results of functions, or stored as components of data structures. The same languages o#er parametric polymorphism, which allows the use of values that behave uniformly over a range of di#erent types. But the combination of these features is not supported--- polymorphic values are not first-class. This restriction is sometimes attributed to the dependence of such languages on type inference, in contrast to more expressive, explicitly typed languages, like System F, that do support first-class polymorphism. This paper uses relationships between types and logic to develop a type system, FCP, that supports first-class polymorphism, type inference, and also first-class abstract datatypes. The immediate result is a more expressive language, but there are also long term implications for language design. 1