Discussion on: ReasonML for production React Apps? 🤔 (Part 2)

View post

Replies for: Nice post. Are you planning to cover a smoother JSON decode story in your next article? I don't want readers to go away with the impression that JS...

As in a simpler example or a simpler way of achieving the same thing? If it's the latter, and you know how to do it, I'd love to receive some guidance and I refactor the article's example.

Yawar Amin • May 4 '19 • Edited

Check out github.com/ryb73/ppx_decco , it's a JSON codec PPX. It should work something like this:

[@decco] type error = {message: string};
[@decco] type response = {
  // other response fields,
  error: option(error),
};

Now it can decode a Js.Json.t that you got from Fetch.Response.json:

|> then_(response =>
  switch (response_decode(response)) {
  | Belt.Result.Ok({error: Some({message})}) =>
    reject(PostError(message))
  | response =>
    resolve(response))

This will resolve with a Js.Promise.t(Belt.Result.t(response, error)) meaning the service didn't return an error message but the JSON may or may not have parsed successfully, or reject with your exception containing the service error message.

Seif Ghezala 🇩🇿 • May 4 '19 • Edited

This syntax looks much cleaner 🤩. I will give it a try soon with the Request module and comment about it in this part of the article. Thank you!