Codehound: Helping Instructors Track Pedagogical Code Dependencies in Course Materials

Abstract

Instructors of programming courses must manage a variety of pedagogical dependencies in their teaching materials. For instance, updating the code used in a single lesson can require cascading changes to other lessons in the course. Currently, they must manually maintain these dependencies across many files, which is tedious and error-prone. To help instructors track pedagogical code dependencies, we created a system called Codehound that uses static analysis to automatically detect where functions are introduced and reused through an entire course. To show how Codehound can be used, we present three usage scenarios inspired by our own experiences teaching large data science courses. These scenarios demonstrate how Codehound can help instructors create new content, collaborate with staff to refactor existing content, and estimate the cost of future course changes.