Haste vs Fay

Haste was designed to allow both the client and server to be written as parts of the same, type-safe application. This is in stark contrast to most other options, where the client and server are considered two separate entities, resulting in extra manual validation code and more chances for type errors.

Haste supports the Haskell 2010 standard except for Template Haskell as well as most GHC extensions.

Haste lets you write client and server as a single program, automatically generating code for the server as well as the client, giving you full type safety even across the Internet.

Fay produces smaller output than pure Haskell compilers such as GHCJS; It does not need to include the whole Haskell runtime, as it drops support for features such as multi-threading, giving it fewer dependencies.

As with UHC, the FFI to Javascript works with printf-style format:

max = ffi "Math.round(%1,%2)"

This can simplify code needed to make calls to methods on objects, in contrast to e.g., Purescript's FFI, which requires that methods be wrapped in Javascript. Similarly to UHC, Fay also supports the use of %*, for javascript functions with arbitrary numbers of parameters, such as concat, though they must expose an explicit number of parameters to Fay.

Fay is available on Cabal, as are Fay packages, so getting up and running is as simple as typing 'cabal install'. Happstack, Snap, and Yesod packages are available on Hackage, as are bindings for JQuery and Backbone.

Since Fay is a subset of Haskell - Lazy, statically typed, and pure by default. There's no new syntax to learn, and no surprises when it comes to the semantics of your code. This extends into function names as well - Fay programmers can use familiar functions such as putStrLn to output to the console, rather than Javascript-specific versions.