Abstract

Generic programming is a technique for writing code that is reusable for many different datatypes. This is an effective technique for eliminating boilerplate code, and many programming languages have at least some support for writing generic programs. However, most languages fall flat in that either (1) their generic programming capabilities are limited to a few built-in (i.e., magical) functions, or (2) rely on a heavyweight macro system that makes it difficult to write generic programs in a cheap and cheerful way.

In this talk, I present an alternative generic programming mechanism found in the Glasgow Haskell Compiler (GHC), referred to as GHC Generics. GHC’s approach greatly simplifies the effort needed to write generic programs and requires almost no compiler magic. We explore how GHC Generics makes it possible to write polytypic code by viewing any algebraic data type as a polynomial functor. We also discuss the tradeoffs involved when using GHC Generics as opposed to other generic programming techniques.