We present a sound type-based `usage analysis' for a realistic lazy functional language. Accurate information on the usage of program subexpressions in a lazy functional language permits a compiler to perform a number of useful optimisations. However, existing analyses are either ad-hoc and approximate, or defined over restricted languages. Our work extends the Once Upon A Type system of Turner, Mossin, and Wadler (FPCA'95). Firstly, we add type polymorphism, an essential feature of typed functional programming languages. Secondly, we include general Haskell-style user-defined algebraic data types. Thirdly, we explain and solve the `poisoning problem', which causes the earlier analysis to yield poor results. Interesting design choices turn up in each of these areas. Our analysis is sound with respect to a Launchbury-style operational semantics, and it is straightforward to implement. Good results have been obtained from a prototype implementation, and we are currently integrating the system into the Glasgow Haskell Compiler.

Abstract not found

The context for this paper is functional computation by graph reduction. Our overall aim is more efficient use of memory. The specific topic is the detection of dormant cells in the live graph --- those retained in heap memory though not actually playing a useful role in computation. We describe a profiler that can identify heap consumption by such `useless' cells. Unlike heap profilers based on traversals of the live heap, this profiler works by examining cells postmortem. The new profiler has revealed a surprisingly large proportion of `useless' cells, even in some programs that previously seemed space-efficient such as the boot-strapping Haskell compiler nhc. 1 Introduction A typical computation by graph reduction involves a large and changing population of heap-memory cells. Taking a census of this population at regular intervals can be very instructive, both for functional programmers and for functional-language implementors. A heap profiler [RW93] records population counts for ...

In this paper we present a non-deterministic call-by-need (untyped) lambda calculus nd with a constant choice and a let-syntax that models sharing. Our main result is that nd has the nice operational properties of the standard lambda calculus: confluence on sets of expressions, and normal order reduction is sufficient to reach head normal form. Using a strong contextual equivalence we show correctness of several program transformations. In particular of lambdalifting using deterministic maximal free expressions. These results show that nd is a new and also natural combination of non-determinism and lambda-calculus, which has a lot of opportunities for parallel evaluation. An intended application of nd is as a foundation for compiling lazy functional programming languages with I/O based on direct calls. The set of correct program transformations can be rigorously distinguished from non-correct ones. All program transformations are permitted with the slight exception that for transformations like common subexpression elimination and lambda-lifting with maximal free expressions the involved subexpressions have to be deterministic ones.

In Part I [5], we proposed an approach to formally describe and compare functional languages implementations. We focused on call-by-value and described well-known compilers for strict languages. Here, we complete our exploration of the design space of implementations by studying call-by-name, call-by-need and graph reduction. We express the whole compilation process as a succession of program transformations in a common framework. At each step, different transformations model fundamental choices or optimizations. We describe and compare the diverse alternatives for the compilation of the call-byname strategy in both environment and graph-based models. The different options for the compilation of b-reduction described in [5] can be applied here as well. Instead, we describe other possibilities specific to graph reduction. Call-by-need is nothing but call-by-name with redex sharing and update. We present how sharing can be expressed in our framework and we describe different...

Traditional database management systems perform updates-in-place and use logs and periodic checkpointing to efficiently achieve atomicity and durability. In this Thesis we shall present a different method, Shades, for achieving atomicity and durability using a copy-on-write policy instead of updates-in-place. We shall also present index structures and the implementation of Shines, a persistent functional programming language, built on top of Shades. Shades includes real-time generational garbage collection. Real-timeness is achieved by collecting only a small part, a generation, of the database at a time. Contrary to previously presented persistent garbage collection algorithms, Shades has no need to maintain metadata (remembered sets) of intra-generation pointers on disk since the metadata can be reconstructed during recovery. This considerably reduces the amount of disk writing. In conjunction with aggressive commit grouping, efficient index structures, a design specialized to a main memory environment, and a carefully crafted implementation of Shines, we have achieved surprisingly high performance, handsomely beating commercial database management systems.

Existing functional languages appear not to be suitable for implementing systems which are inherently concurrent, such as operating system environments. Adaptations to functional languages developed to support such applications have in the past always involved the introduction of non-determinism.

: We introduce a unified framework to describe, relate, compare and classify functional language implementations. The compilation process is expressed as a succession of program transformations in the common framework. At each step, different transformations model fundamental choices. A benefit of this approach is to structure and decompose the implementation process. The correctness proofs can be tackled independently for each step and amount to proving program transformations in the functional world. This approach also paves the way to formal comparisons by making it possible to estimate the complexity of individual transformations or compositions of them. Our study aims at covering the whole known design space of sequential functional languages implementations. In particular, we consider call-by-value, call-by-name and call-by-need reduction strategies as well as environment and graph-based implementations. We describe for each compilation step the diverse alternatives as program tr...

Current implementation techniques for functional languages differ considerably from those for logic languages. This complicates the development of flexible and efficient abstract machines that can be used for the compilation of declarative languages combining concepts of functional and logic programming. We propose an abstract machine, called the JUMP-machine, which systematically integrates the operational concepts needed to implement the functional and logic programming paradigm. The use of a tagless representation for heap objects, which originates from the Spineless Tagless G-machine, supports the integration of different concepts. In this paper, we provide a functional logic kernel language and show how to translate it into the abstract machine language of the JUMP-machine. Furthermore, we define the operational semantics of the machine language formally and discuss the mapping of the abstract machine to concrete machine architectures. We tested the approach by writing a compiler for the functional logic language GTML. The obtained performance results indicate that the proposed method allows to implement functional logic languages efficiently.

Abstract. This chapter describes Hume: a functionally-based language for programming with bounded resource usage, including time and space properties. The purpose of the Hume language design is to explore the expressibility/costability spectrum in resource-constrained systems, such as real-time embedded or control systems. It is unusual in being based on a combination of λ-calculus and finite state machine notions, rather than the more usual propositional logic, or flat finite-state-machine models. The use of a strict, purely functional programming notation allows the construction of a strong cost model for expressions, which can then be embedded into a simple cost model for processes. In this chapter, we introduce Hume, describe the Hume Abstract Machine implementation, and show how a high-level cost model can be constructed that relates costs from the abstract machine to Hume source programs. We illustrate our approach with an example adapted from the literature: a simple vending machine controller. 1