Trellys: Dependently-typed Language Design

My research focuses on the design of statically-typed programming languages. Static type systems are a popular, cost-effective form of lightweight program verification. They provide a tractable and modular way for programmers to express properties that can be mechanically checked by the compiler. As a result, the compiler can rule out a wide variety of errors and provide more information to refactoring and development tools. For example, systems written with type-safe languages cannot be compromised by buffer overruns if all array accesses are statically proven safe. Furthermore, programmers can modify their code with the assurance that they have not violated critical safety properties. I explore these designs in the context of functional programming languages, such as Haskell and ML. Functional programming languages are an ideal context for type system research; they excel in their capabilities for static reasoning. However, there is need for improvement. Some programming idioms must be ruled out simply because they cannot be shown to be sound by existing type systems. To overcome these limitations, my work investigates type system features in the context of both new languages and existing ones, and evaluates those designs with respect to both theory and practice.