A Framework for Generating Diverse Haskell-I/O Exercise Tasks

Abstract

We present the design of a framework to describe parametrized exercise tasks on Haskell-I/O programming. Parametrized tasks can be instantiated randomly to quickly generate different instances of a task. Such automatic task generation is useful in many different ways. Manual task creation can be a time-consuming process, so formulating a task design once and then automatically generating different variations can save valuable time for the educator. The descriptions of tasks also serve as easy to understand documentation and can be reused in new task designs. On the student’s side automatic task generation, together with an automated assessment system, enables practicing on as many fresh exercise tasks as needed. Students can also each be given a slightly different version of tasks, reducing issues regarding plagiarism arising naturally in an e-learning environment. Our task generation is centered around a specification language for I/O behavior we developed in earlier work. The task generation framework, an embedded domain specific language in Haskell, provides powerful primitives for the creation of various artifacts from specifications, including program code. We do not go into detail on the technical realization of these primitives. Our focus is on showcasing how such artifacts can be used as an alternative to the verbal description of requirements for different types of programming exercise tasks. By doing so, we are able to automatically generate a diverse range of task designs.