To avoid compilation errors it is desirable to verify that a compiler is type correct—i.e., given well-typed source code, it always outputs well-typed target code. This can be done intrinsically by implementing it as a function in a dependently typed programming language, such as Agda. This function manipulates data types of well-typed source and target programs, and is therefore type correct by construction. A key challenge in implementing an intrinsically typed compiler is the representation of labels in bytecode. Because label names are global, bytecode typing appears to be inherently a non-compositional, whole-program property. The individual operations of the compiler do not preserve this property, which requires the programmer to reason about labels, which spoils the compiler definition with proof terms. In this paper, we address this problem using a new nameless and co-contextual representation of typed global label binding, which is compositional. Our key idea is to use linearity to ensure that all labels are defined exactly once. To write concise compilers that manipulate programs in our representation, we develop a linear, dependently typed, shallowly embedded language in Agda, based on separation logic. We show that this language enables the concise specification and implementation of intrinsically typed operations on bytecode, culminating in an intrinsically typed compiler for a language with structured control-flow.

There is a large gap between the specification of type systems and the implementation of their type checkers, which impedes reasoning about the soundness of the type checker with respect to the specification. A vision to close this gap is to automatically obtain type checkers from declarative programming language specifications. This moves the burden of proving correctness from a case-by-case basis for concrete languages to a single correctness proof for the specification language. This vision is obstructed by an aspect common to all programming languages: name resolution. Naming and scoping are pervasive and complex aspects of the static semantics of programming languages. Implementations of type checkers for languages with name binding features such as modules, imports, classes, and inheritance interleave collection of binding information (i.e., declarations, scoping structure, and imports) and querying that information. This requires scheduling those two aspects in such a way that query answers are stable—i.e., they are computed only after all relevant binding structure has been collected. Type checkers for concrete languages accomplish stability using language-specific knowledge about the type system. In this paper we give a language-independent characterization of necessary and sufficient conditions to guarantee stability of name and type queries during type checking in terms of critical edges in an incomplete scope graph. We use critical edges to give a formal small-step operational semantics to a declarative specification language for type systems, that achieves soundness by delaying queries that may depend on missing information. This yields type checkers for the specified languages that are sound by construction—i.e., they schedule queries so that the answers are stable, and only accept programs that are name- and type-correct according to the declarative language specification. We implement this approach, and evaluate it against specifications of a small module and record language, as well as subsets of Java and Scala.

Context: Compilation time is an important factor in the adaptability of a software project. Fast recompilation enables cheap experimentation with changes to a project, as those changes can be tested quickly. Separate and incremental compilation has been a topic of interest for a long time to facilitate fast recompilation. Inquiry: Despite the benefits of an incremental compiler, such compilers are usually not the default. This is because incrementalization requires cross-cutting, complicated, and error-prone techniques such as dependency tracking, caching, cache invalidation, and change detection. Especially in compilers for languages with cross-module definitions and integration, correctly and efficiently implementing an incremental compiler can be a challenge. Retrofitting incrementality into a compiler is even harder. We address this problem by developing a compiler design approach that reuses parts of an existing non-incremental compiler to lower the cost of building an incremental compiler. It also gives an intuition into compiling difficult-to-incrementalize language features through staging. Approach: We use the compiler design approach presented in this paper to develop an incremental com- piler for the Stratego term-rewriting language. This language has a set of features that at first glance look incompatible with incremental compilation. Therefore, we treat Stratego as our critical case to demonstrate the approach on. We show how this approach decomposes the original compiler and has a solution to com- pile Stratego incrementally. The key idea on which we build our incremental compiler is to internally use an incremental build system to wire together the components we extract from the original compiler. Knowledge: The resulting compiler is already in use as a replacement of the original whole-program compiler. We find that the incremental build system inside the compiler is a crucial component of our approach. This allows a compiler writer to think in multiple steps of compilation, and combine that into a incremental compiler almost effortlessly. Normally, separate compilation à la C is facilitated by an external build system, where the programmer is responsible for managing dependencies between files. We reuse an existing sound and optimal incremental build system, and integrate its dependency tracking into the compiler. Grounding: The incremental compiler for Stratego is available as an artefact along with this article. We evaluate it on a large Stratego project to test its performance. The benchmark replays edits to the Stratego project from version control. These benchmarks are part of the artefact, packaged as a virtual machine image for easy reproducibility. Importance: Although we demonstrate our design approach on the Stratego programming language, we also describe it generally throughout this paper. Many currently used programming languages have a compiler that is much slower than necessary. Our design provides an approach to change this, by reusing an existing compiler and making it incremental within a reasonable amount of time.