Partial evaluation reconciles generality with efficiency by providing automatic specialization and optimization of programs. This book covers the entire field of partial evaluation; provides simple and complete algorithms; and demonstrates that specialization can increase efficiency.

As the complexity of software increases, researchers and practicioners continue to seek better techniques for engineering the construction of evolution of software. Partial evaluation is an attractive technology for modern software construction since it provides automatic tools for software specialization and is based on rigorous semantic foundations. This book is based on a school held at DIKU Copenhagen, Denmark in summer 1998 during which leading researchers summarized the state of the art in partial evaluation. The lectures presented survey the foundations of partial evaluation in a clear and rigorous manner and practically introduce several existing partial evaluators with numerous examples. The second part of the book is devoted to more sophisticated theoretical aspects, advances systems and applications, and highlights open problems and challenges. The book is ideally suited for advanced courses and for self study.

Program generation holds the promise of helping to bridge the gap between application-level problem solutions and efficient implementations at the level of today's source programs as written in C or Java. Thus, program generation can substantially contribute to reducing production cost and time-to-market in future software production, while improving the quality and stability of the product. This book is about domain-specific program generation; it is the outcome of a Dagstuhl seminar on the topic held in March 2003. After an introductory preface by the volume editors, the 18 carefully reviewed revised full papers presented are organized into topical sections on - surveys of domain-specific programming technologies - domain-specific programming languages - tool support for program generation - domain-specific techniques for program optimization

This book uses a functional programming language (F#) as a metalanguage to present all concepts and examples, and thus has an operational flavour, enabling practical experiments and exercises. It includes basic concepts such as abstract syntax, interpretation, stack machines, compilation, type checking, garbage collection, and real machine code. Also included are more advanced topics on polymorphic types, type inference using unification, co- and contravariant types, continuations, and backwards code generation with on-the-fly peephole optimization. This second edition includes two new chapters. One describes compilation and type checking of a full functional language, tying together the previous chapters. The other describes how to compile a C subset to real (x86) hardware, as a smooth extension of the previously presented compilers.The examples present several interpreters and compilers for toy languages, including compilers for a small but usable subset of C, abstract machines, a garbage collector, and ML-style polymorphic type inference. Each chapter has exercises. Programming Language Concepts covers practical construction of lexers and parsers, but not regular expressions, automata and grammars, which are well covered already. It discusses the design and technology of Java and C# to strengthen students’ understanding of these widely used languages.

Computability and complexity theory should be of central concern to practitioners as well as theorists. Unfortunately, however, the field is known for its impenetrability. Neil Jones's goal as an educator and author is to build a bridge between computability and complexity theory and other areas of computer science, especially programming. In a shift away from the Turing machine- and G�del number-oriented classical approaches, Jones uses concepts familiar from programming languages to make computability and complexity more accessible to computer scientists and more applicable to practical programming problems. According to Jones, the fields of computability and complexity theory, as well as programming languages and semantics, have a great deal to offer each other. Computability and complexity theory have a breadth, depth, and generality not often seen in programming languages. The programming language community, meanwhile, has a firm grasp of algorithm design, presentation, and implementation. In addition, programming languages sometimes provide computational models that are more realistic in certain crucial aspects than traditional models. New results in the book include a proof that constant time factors do matter for its programming-oriented model of computation. (In contrast, Turing machines have a counterintuitive "constant speedup" property: that almost any program can be made to run faster, by any amount. Its proof involves techniques irrelevant to practice.) Further results include simple characterizations in programming terms of the central complexity classes PTIME and LOGSPACE, and a new approach to complete problems for NLOGSPACE, PTIME, NPTIME, and PSPACE, uniformly based on Boolean programs. Foundations of Computing series