Effective Haskell

Procedural Shell, Functional Core

One of the most common problems that people encounter when they first start trying to write larger Haskell is programs is understanding how to handle IO architecturally. Even when you understand the mechanics of how IO works in the language, it can be hard to figure out how to apply those ideas when you are building a complete program. The problem is that when you are coming from a language where you can do IO at any time, you get accustomed to writing programs that do IO as needed. If you visualize your program as a call graph, any individual function in the call graph might do a little bit of IO because it needs to look up a value from the environment, make a network request, or write a file. In many cases, ...