Chocopy: A programming language for compilers courses

Abstract

ChocoPy is a programming language designed for teaching an undergraduate course on programming languages and compilers. ChocoPy is a restricted subset of Python 3.6, using static type annotations to enforce compile-time type safety. ChocoPy is fully specified using formal grammar, typing rules, and operational semantics. Valid ChocoPy programs can be executed in a standard Python interpreter, producing results consistent with ChocoPy semantics. A major component of CS164 at UC Berkeley is the project: students develop a full compiler for ChocoPy, targeting RISC-V, in about twelve weeks. In other exercises, students extend the syntax, type system, and formal semantics to support additional features of Python. In this paper, we outline (1) the motivations for creating the ChocoPy project, (2) salient features of the language, (3) the resources provided to students to develop their compiler, (4) some insights gained from teaching two semesters of ChocoPy-based courses by different instructors. Our assignment resources are available for re-use by other instructors and institutions.