Functional reactive programming (FRP) is a powerful paradigm for writing code that reacts to change, and has been gaining significant traction as a way to implement GUI applications with minimal state.
Haskell has a few existing FRP libraries that are commonly used for GUI programming; however, all of them currently target cross-platform widget toolkits, like GTK+ or wxWidgets. Native frameworks (like Cocoa or WPF) have remained highly imperative, and out of place in the FRP model.
This talk introduces RxHaskell (http://hackage.haskell.org/package/RxHaskell) – a library for functional reactive programming inspired by Microsoft's Reactive Extensions for .NET (http://msdn.microsoft.com/en-us/library/hh242985(v=VS.103).aspx) and designed specifically for integration with native frameworks.
Rx and RxHaskell simplify the traditional FRP model by omitting continuous behaviors, and discarding values after they've been processed (which enables switching without time leaks). These changes make it easier to interleave impure code.
And because native frameworks often require UI changes to be made on a dedicated OS thread, RxHaskell provides a "scheduler" abstraction which uses Haskell's type system to offer static guarantees about concurrency. Functions can be annotated as running on the main thread or in the background, and an action of one type cannot be directly executed from the other, preventing some classes of threading error.
We'll also take a brief look at how RxHaskell can be bridged to ReactiveCocoa, an Rx implementation in Objective-C, to break down the barriers between two very different programming languages, and support amazing native development without giving way to completely imperative programming.