This package helps you convert between Elm values and JSON values.

This package is usually used alongside elm/http to talk to servers or ports to talk to JavaScript.

Have you seen this causes of death table? Did you know that in 2002, war accounted for 0.3% of global deaths whereas road traffic accidents accounted for 2.09% and diarrhea accounted for 3.15%?

The table is interesting, but say we want to visualize this data in a nicer way. We will need some way to get the cause-of-death data from our server, so we create encoders and decoders:

module Cause exposing (Cause, encode, decoder)

import Json.Decode as D
import Json.Encode as E


-- CAUSE OF DEATH

type alias Cause =
  { name : String
  , percent : Float
  , per100k : Float
  }


-- ENCODE

encode : Cause -> E.Value
encode cause =
  E.object
    [ ("name", E.string cause.name)
    , ("percent", E.float cause.percent)
    , ("per100k", E.float cause.per100k)
    ]


-- DECODER

decoder : D.Decoder Cause
decoder =
  D.map3 Cause
    (D.field "name" D.string)
    (D.field "percent" D.float)
    (D.field "per100k" D.float)

Now in some other code we can use Cause.encode and Cause.decoder as building blocks. So if we want to decode a list of causes, saying Decode.list Cause.decoder will handle it!

Point is, the goal should be:

1. Make small JSON decoders and encoders.
2. Snap together these building blocks as needed.

So say you decide to make the name field more precise. Instead of a String, you want to use codes from the International Classification of Diseases recommended by the World Health Organization. These codes are used in a lot of mortality data sets. So it may make sense to make a separate IcdCode module with its own IcdCode.encode and IcdCode.decoder that ensure you are working with valid codes. From there, you can use them as building blocks in the Cause module!

It is easy to get focused on how to optimize the use of JSON, but I think this is missing the bigger picture. Instead, I would like to head towards this vision of data interchange.