The G-machine (Johnsson, 1987; Peyton Jones, 1987) is a compiled graph reduction machine for lazy functional languages. The G-machine compiler contains many optimizations to improve performance. One set of such optimizations is designed to improve the performance of tail recursive functions. Unfortunately, the abstract machine is subject to a space leak—objects are unnecessarily preserved by the garbage collector. This paper analyses why a particular form of space leak occurs in the G-machine, and presents some ideas for fixing this problem. This phenomena in other abstract machines is also examined briefly.

Chris Okasaki showed how to implement red-black trees in a functional programming language. Ralf Hinze incorporated even the invariants of such data structures into their types, using higher-order nested datatypes. We show how one can achieve something very similar without the usual performance penalty of such types, by combining the features of nested datatypes, phantom types and existential type variables.

This paper proves correctness of Nöcker's method of strictness analysis, implemented in the Clean compiler, which is an effective way for strictness analysis in lazy functional languages based on their operational semantics. We improve upon the work Clark, Hankin and Hunt did on the correctness of the abstract reduction rules in two aspects. Our correctness proof is based on a functional core language and a contextual semantics, thus proving a wider range of strictness-based optimizations as correct, and our method fully considers the cycle detection rules, which contribute to the strength of Nöcker's strictness analysis.

Our algorithm SAL is a reformulation of Nöcker's strictness analysis algorithm in a functional core language LR. This is a higher order call-by-need lambda calculus with case, constructors, letrec, and seq, which is extended during strictness analysis by set constants like Top or Inf, denoting sets of expressions, which indicate different evaluation demands. It is also possible to define new set constants by recursive equations with a greatest fixpoint semantics. The operational semantics of LR is a small-step semantics. Equality of expressions is defined by a contextual semantics that observes termination of expressions. Basically, SAL is a nontermination checker. The proof of its correctness and hence of Nöcker's strictness analysis is based mainly on an exact analysis of the lengths of evaluations, i.e., normal-order reduction sequences to WHNF. The main measure being the number of “essential” reductions in evaluations.

Our tools and results provide new insights into call-by-need lambda calculi, the role of sharing in functional programming languages, and into strictness analysis in general. The correctness result provides a foundation for Nöcker's strictness analysis in Clean, and also for its use in Haskell.

The Theory of Contexts is a type-theoretic axiomatization aiming to give a metalogical account of the fundamental notions of variable and context as they appear in Higher Order Abstract Syntax. In this paper, we prove that this theory is consistent by building a model based on functor categories. By means of a suitable notion of forcing, we prove that this model validates Classical Higher Order Logic, the Theory of Contexts, and also (parametrised) structural induction and recursion principles over contexts. Our approach, which we present in full detail, should also be useful for reasoning on other models based on functor categories. Moreover, the construction could also be adopted, and possibly generalized, for validating other theories of names and binders.

Mogensen has exhibited a very compact partial evaluator for the pure lambda calculus, using binding-time analysis followed by specialization. We give a correctness criterion for this partial evaluator and prove its correctness relative to this specification. We show that the conventional properties of partial evaluators, such as the Futamura projections, are consequences of this specification. By considering both a flow analysis and the transformation it justifies together, this proof suggests a framework for incorporating flow analyses into verified compilers.

First-order unification algorithms (Robinson, 1965) are traditionally implemented via general recursion, with separate proofs for partial correctness and termination. The latter tends to involve counting the number of unsolved variables and showing that this total decreases each time a substitution enlarges the terms. There are many such proofs in the literature (Manna & Waldinger, 1981; Paulson, 1985; Coen, 1992; Rouyer, 1992; Jaume, 1997; Bove, 1999). This paper shows how a dependent type can relate terms to the set of variables over which they are constructed. As a consequence, first-order unification becomes a structurally recursive program, and a termination proof is no longer required. Both the program and its correctness proof have been checked using the proof assistant LEGO (Luo & Pollack, 1992; McBride, 1999).

We present a straightforward, sound, Hindley–Milner polymorphic type system for algebraic effects and handlers in a call-by-value calculus, which, to our surprise, allows type variable generalisation of arbitrary computations, and not just values. We first recall that the soundness of unrestricted call-by-value Hindley–Milner polymorphism is known to fail in the presence of computational effects such as reference cells and continuations, and that many programming examples can be recast to use effect handlers instead of these effects. After presenting the calculus and its soundness proof, formalised in Twelf, we analyse the expressive power of effect handlers with respect to state effects. We conjecture handlers alone cannot express reference cells, but show they can simulate dynamically scoped state, establishing that dynamic binding also does not need a value restriction.

Pattern matching in modern functional programming languages is tied to the representation of data. Unfortunately, this is incompatible with the philosophy of abstract data types.

Two proposals have been made to generalize pattern matching to a broader class of types. The laws mechanism of Miranda allows pattern matching with non-free algebraic data types. More recently, Wadler proposed the concept of views as a more general solution, making it possible to define arbitrary mappings between a physical implementation and a view supporting pattern matching. Originally, it was intended to include views in the new standard lazy functional programming language Haskell.

Laws and views each offer important advantages, particularly with respect to data abstraction. However, if not used with great care, they also introduce serious problems in equational reasoning. As a result, laws have been removed from Miranda and views were not included in the final version of Haskell.

We propose a third approach which unifies the laws and views mechanisms while avoiding their problems. Philosophically, we view pattern matching as a bundling of case recognition and component selection functions instead of a method for inverting data construction. This can be achieved by removing the implied equivalence between data constructors and pattern constructors. In practice, we allow automatic mapping into a view but not out of the view. We show that equational reasoning can still be used with the resulting system. In fact, equational reasoning is easier, since there are fewer hidden traps.

Pure type systems make use of domain-full λ-abstractions λx:D.M. We present a variant of pure type systems, which we call domain-free pure type systems, with domain-free λ-abstractions λx.M. Domain-free pure type systems have a number of advantages over both pure type systems and so-called type assignment systems (they also have some disadvantages), and have been used in theoretical developments as well as in implementations of proof-assistants. We study the basic properties of domain-free pure type systems, establish their formal relationship with pure type systems and type assignment systems, and give a number of applications of these correspondences.

Polymorphic types are labels classifying both (a) defined components in a library and (b) contexts of free variables in partially written programs. It is proposed to help programmers make better use of software libraries by providing a system that, given (b), identifies candidates from (a) with matching types. Assuming at first that matching means unifying (i.e. having a common instance), efficient ways of implementing such a retrieval system are discussed and its likely effectiveness based on a quantitative study of currently available libraries is indicated. The applicative instance relation between types, which captures some intuitions about generalization/specialization is then introduced, and its use as the basis of a more flexible system is discussed.

We introduce the arrow calculus, a metalanguage for manipulating Hughes's arrows with close relations both to Moggi's metalanguage for monads and to Paterson's arrow notation. Arrows are classically defined by extending lambda calculus with three constructs satisfying nine (somewhat idiosyncratic) laws; in contrast, the arrow calculus adds four constructs satisfying five laws (which fit two well-known patterns). The five laws were previously known to be sound; we show that they are also complete, and hence that the five laws may replace the nine.

We study the type inference problem for a system with type classes as in the functional programming language Haskell. Type classes are an extension of ML-style polymorphism with overloading. We generalize Milner's work on polymorphism by introducing a separate context constraining the type variables in a typing judgement. This leads to simple type inference systems and algorithms which closely resemble those for ML. In particular, we present a new unification algorithm which is an extension of syntactic unification with constraint solving. The existence of principal types follows from an analysis of this unification algorithm.

In this paper we present the algebraic-λ-cube, an extension of Barendregt's λ-cube with first- and higher-order algebraic rewriting. We show that strong normalization is a modular property of all the systems in the algebraic-λ-cube, provided that the first-order rewrite rules are non-duplicating and the higher-order rules satisfy the general schema of Jouannaud and Okada. We also prove that local confluence is a modular property of all the systems in the algebraic-λ-cube, provided that the higher-order rules do not introduce critical pairs. This property and the strong normalization result imply the modularity of confluence.

Family polymorphism has been proposed for object-oriented languages as a solution to supporting reusable yet type-safe mutually recursive classes. A key idea of family polymorphism is the notion of families, which are used to group mutually recursive classes. In the original proposal, due to the design decision that families are represented by objects, dependent types had to be introduced, resulting in a rather complex type system. In this article, we propose a simpler solution of lightweight family polymorphism, based on the idea that families are represented by classes rather than by objects. This change makes the type system significantly simpler without losing much expressive power of the language. Moreover, “family-polymorphic” methods now take a form of parametric methods; thus, it is easy to apply method type argument inference as in Java 5.0. To rigorously show that our approach is safe, we formalize the set of language features on top of Featherweight Java and prove that the type system is sound. An algorithm for type inference for family-polymorphic method invocations is also formalized and proved to be correct. Finally, a formal translation by erasure to Featherweight Java is presented; it is proved to preserve typing and execution results, showing that our new language features can be implemented in Java by simply extending the compiler.

Thirty-five years ago, thunks were used to simulate call-by-name under call-by-value in Algol 60. Twenty years ago, Plotkin presented continuation-based simulations of call-by-name under call-by-value and vice versa in the λ-calculus. We connect all three of these classical simulations by factorizing the continuation-based call-by-name simulation [Cscr ]n with a thunk-based call-by-name simulation [Tscr ] followed by the continuation-based call-by-value simulation [Cscr ]v, extended to thunks.

formula here

We show that [Tscr ] actually satisfies all of Plotkin's correctness criteria for [Cscr ]n (i.e. his Indifference, Simulation and Translation theorems). Furthermore, most of the correctness theorems for [Cscr ]n can now be seen as simple corollaries of the corresponding theorems for [Cscr ]v and [Tscr ].

We present two implementations of Oppen's pretty-printing algorithm in Haskell that meet the efficiency of Oppen's imperative solution but have a simpler and a clear structure. We start with an implementation that uses lazy evaluation to simulate two co-operating processes. Then we present an implementation that uses higher-order functions for delimited continuations to simulate co-routines with explicit scheduling.

Unification, or two-way pattern matching, is the process of solving an equation involving two first-order terms with variables. Unification is used in type inference in many programming languages and in the execution of logic programs. This means that unification algorithms have to be written over and over again for different term types. Many other functions also make sense for a large class of datatypes; examples are pretty printers, equality checks, maps etc. They can be defined by induction on the structure of user-defined datatypes. Implementations of these functions for different datatypes are closely related to the structure of the datatypes. We call such functions polytypic. This paper describes a unification algorithm parametrised on the type of the terms, and shows how to use polytypism to obtain a unification algorithm that works for all regular term types.

Abstract interpretation is the collective name for a family of semantics-based techniques for compile-time analysis of programs. One of the most costly operations in automating such analyses is the computation of fixed points. The frontiers algorithm is an elegant method, invented by Chris Clack and Simon Peyton Jones, which addresses this issue.

In this article we present a new approach to the frontiers algorithm based on the insight that frontiers represent upper and lower subsets of a function's argument domain. This insight leads to a new formulation of the frontiers algorithm for higher-order functions, which is considerably more concise than previous versions.

We go on to argue that for many functions, especially in the higher-order case, finding fixed points is an intractable problem unless the sizes of the abstract domains are reduced. We show how the semantic machinery of abstract interpretation allows us to place upper and lower bounds on the values of fixed points in large lattices by working within smaller ones.

 Bibliography  241

 Index  243

We give a direct type-theoretic characterization of the basic mechanisms of object-oriented programming, including objects, methods, message passing, and subtyping, by introducing an explicit constructor for object types and suitable introduction, elimination, and equality rules. The resulting abstract framework provides a basis for justifying and comparing previous encodings of objects based on recursive record types (Cardelli, 1984; Cardelli, 1992; Bruce, 1994; Cook et al., 1990; Mitchell, 1990a) and encodings based on existential types (Pierce & Turner, 1994).