Datatype-Generic Programming in Haskell

9th April 2013 in London at The Skills Matter eXchange

This SkillsCast was filmed at Datatype-Generic Programming in Haskell

In the talk, I will explain all about how datatype-generic programming in GHC works, and I will present several examples of its use, including examples of how it already is successfully being used in some prominent Haskell libraries.

Haskell's deriving construct is both wonderful and magical at the same time: it is wonderful, because with just a single line of code, you can obtain functionality such as(de)serialization, comparison and traversal functions for your newly defined datatype; it is magical because all of this happens behind the scenes, by code deeply built into the compiler.

But does it have to be magic? Since recently, the Glasgow Haskell Compiler (GHC) supports a feature that lets you define your own derivable functionality! And it is not difficult to use, either. All you have to do is to understand a little bit about how Haskell datatypes can be uniformly represented using a limited set of simple datatypes. Then you define a class plus instances for this limited set of types, and get all the rest for free.

So-called datatype-generic functions let you properly express your code in terms of specific, user-defined datatypes without paying a price: for your new datatypes, you can still use lots of pre-defined functionality, simply by deriving it.

Datatype-Generic Programming in Haskell

Andres Löh

Andres Löh is a long-time functional programming enthusiast. He has started using Haskell in 1997 while being an undergraduate studying mathematics. He obtained his PhD on datatype-generic programming using Haskell from Utrecht University in 2004. Since then, he has continued to use Haskell in research and practice, including teaching various courses both to students and participants from industry. His main interests, next to datatype-generic programming, are embedded domain-specific languages, unleashing the full power of Haskell's type system, parallelism and applying functional languages to real-life problems. Since 2010, Andres is an independent Haskell consultant and partner at Well-Typed LLP.