# Encode

To convert Output a to WAI Response Linnet employs Encode type class:

import Data.ByteString.Lazy (ByteString)

class Encode (ct :: Symbol) a where
encode :: a -> ByteString


Content-Type of response is encoded as phantom type ct, so compiler always picks up the encoder for the specific Content-Type if it exists. Then, type a is converted as lazy ByteString that represents body of a response.

## Encode SomeException

Linnet doesn't make an assumption on how to encode exceptions occurred during request. It's up to user to decide on representation of exceptions if there should be any. The simplest case is to return an empty body:

{-# LANGUAGE FlexibleInstances     #-}
{-# LANGUAGE MultiParamTypeClasses #-}
{-# LANGUAGE TypeSynonymInstances  #-}

import Control.Exception
import Linnet

instance Encode TextPlain SomeException where
encode _ = mempty


In a nutshell, this type class could be treated as a single point of exceptions handling where specific errors are rendered in a way they should be. Only errors inside of Output are encoded this way. Information on the general error handling is available in the corresponding section.

## Decode

Decode type class is responsible for decoding request's body into some type a:

import Data.ByteString.Lazy (ByteString)

class Decode (ct :: Symbol) a where
decode :: ByteString -> Either LinnetError a


And again, Content-Type of expected request is encoded as phantom type ct to ensure that correct decoder exists in compile-time.