A number of important program rewriting scenarios can be recast as type-directed coercion insertion. These range from more theoretical applications such as coercive subtyping and supporting overloading in type theories, to more practical applications such as integrating static and dynamically typed code using gradual typing, and inlining code to enforce security policies such as access control and provenance tracking. In this paper we give a general theory of typedirected coercion insertion. We specifically explore the inherent tradeoff between expressiveness and ambiguity—the more powerful the strategy for generating coercions, the greater the possibility of several, semantically distinct rewritings for a given program. We consider increasingly powerful coercion generation strategies, work out example applications supported by the increased power (including those mentioned above), and identify the inherent ambiguity problems of each setting, along with various techniques to tame the ambiguities.

Abstract. We present a uniqueness type system that is simpler than both Clean’s uniqueness system and a system we proposed previously. The new type system is straightforward to implement and add to existing compilers, and can easily be extended with advanced features such as higher rank types and impredicativity. We describe our implementation in Morrow, an experimental functional language with both these features. Finally, we prove soundness of the core type system with respect to the call-by-need lambda calculus. 1 Introduction to Uniqueness Typing An important property of pure functional programming languages is referential transparency: the same expression used twice must have the same value twice. This makes equational reasoning possible and aids program analysis, but most languages do not have this property. For example, in the following C fragment,

It is common for compilers to derive the calling convention of a function from its type. Doing so is simple and modular but misses many optimisation opportunities, particularly in lazy, higher-order functional languages with extensive use of currying. We restore the lost opportunities by defining Strict Core, a new intermediate language whose type system makes the missing distinctions: laziness is explicit, and functions take multiple arguments and return multiple results.

Haskell’s multi-parameter type classes, together with functional dependencies, allow the specification of complex type-level operations, and the recent introduction of open type families in GHC makes such type-level programming even more accessible and flexible. But type-level code is special in that its correctness is crucial to the safety of the program; so except in those cases simple enough for the type checker to see trivially that the code is correct (or harmless), type-level programs need to come with their specification and correctness proof. In this article, we propose an extension to Haskell that allows the specification of invariants for type classes and open type families, together with accompanying evidence that those invariants hold. To accommodate the open nature of type classes and type families, the evidence itself needs to be open and every subcase of the proof can be provided independently from the others.

In the presence of ever-changing computer architectures, highquality optimising compiler backends are moving targets that require specialist knowledge and sophisticated algorithms. In this paper, we explore a new backend for the Glasgow Haskell Compiler (GHC) that leverages the Low Level Virtual Machine (LLVM), a new breed of compiler written explicitly for use by other compiler writers, not high-level programmers, that promises to enable outsourcing of low-level and architecture-dependent aspects of code generation. We discuss the conceptual challenges and our backend design. We also provide an extensive quantitative evaluation of the performance of the backend and of the code it produces.

The use of typed intermediate languages can significantly increase the reliability of a compiler. By type-checking the code produced at each transformation stage, one can identify bugs in the compiler that would otherwise be much harder to find. We propose to take the use of types in compilation a step further by verifying that the transformation itself is type correct, in the sense that it is impossible that it produces an ill typed term given a well typed term as input. We base our approach on higher-order abstract syntax (HOAS), a representation of programs where variables in the object language are represented by meta-variables. We use a representation that accounts for the object language’s type system using generalized algebraic data types (GADTs). In this way, the full binding and type structure of the object language is exposed to the host language’s type system. In this setting we encode a type preservation property of a CPS conversion in Haskell’s type system, using witnesses of a type correctness proof encoded as values in a GADT.

We propose an Haskell-like language with the goal of unifying type classes and generalized algebraic datatypes (GADTs) into a single class construct. We treat classes as first-class types and we use objects (instead of type class instances and data constructors) to define the values of those classes. We recover the ability to define functions by pattern matching by using sealed classes. The resulting language is simple and intuitive and it can be used to define, with similar convenience, the same programs that we would define in Haskell. Furthermore, unlike Haskell, dictionaries (or objects) can be explicitly (as well as implicitly) passed to functions and we can program in a simple object-oriented style directly. 1.

This paper presents AURA, a programming language for access control that treats ordinary programming constructs (e.g., integers and recursive functions) and authorization logic constructs (e.g., principals and access control policies) in a uniform way. AURA is based on polymorphic DCC and uses dependent types to permit assertions that refer directly to AURA values while keeping computation out of the assertion level to ensure tractability. The main technical results of this paper include fully mechanically verified proofs of the decidability and soundness for AURA’s type system, and a prototype typechecker and interpreter. 1.

We show how the binary encoding and decoding of typed data and typed programs can be understood, programmed, and verified with the help of question-answer games. The encoding of a value is determined by the yes/no answers to a sequence of questions about that value; conversely, decoding is the interpretation of binary data as answers to the same question scheme. We introduce a general framework for writing and verifying gamebased codecs. We present games for structured, recursive, polymorphic, and indexed types, building up to a representation of well-typed terms in the simply-typed λ-calculus. The framework makes novel use of isomorphisms between types in the definition of games. The definition of isomorphisms together with additional simple properties make it easy to prove that codecs derived from games never encode two distinct values using the same code, never decode two codes to the same value, and interpret any bit sequence as a valid code for a value or as a prefix of a valid code.

Type Checking with Open Type Functions We report on an extension of Haskell with open type-level functions and equality constraints that unifies earlier work on GADTs, functional dependencies, and associated types. The contribution of the paper is that we identify and characterise the key technical challenge of entailment checking; and we give a novel, decidable, sound, and complete algorithm to solve it, together with some practically-important variants. Our system is implemented in GHC, and is already in active use.