Concept: JSON Decoders about elm

[mpizenberg]

Maybe encoding a request to the github api (which is well known), and decoding a response?

Good idea to work on a concrete real-life thing. We'd have to rebuild the JS Value ourselves directly in the test though as obviously there is no outside world access in tests file.

[jiegillet]

Oh I still intend to keep rest-api, and I don't think having a difficult exercise is inherently a problem (I love those as a student), I'm just looking for another one that could help practicing decoders.
I don't think that rest-api is THAT complicated to understand if you read the description and the tests carefully though (I'll still try to make it clearer).

[jiegillet]

Creating a proper new practice exercise is way more work than I'm willing to do!
I guess we'll need a really good concept exercise :)

[jiegillet]

@ceddlyburge related to our discussion on #621, I went through the exercises that we don't have implemented yet (bin/configlet info) to look for other exercises that we could use to practice JSON decoding, but hopefully a bit simpler than rest-api.

Basically no other is meant to practice JSON, so we would have to stretch some scopes.

I found two that may be interesting: pov and zipper. Those two exercises involve trees, pov is bout rerooting trees and zipper traversing trees. I say that they are interesting because parsing trees is fun, all the other exercises have very standard, predictable inputs.
Unfortunately, these 2 are not really easy, so it doesn't help us that much.

If we are OK with very simple decoding, maybe we could use affine-cipher.

Best for last: bank-account could work as well, transforming the string instructions into meaningful types.

What do you think?

[ceddlyburge]

I think the api concept is still good, its just that it is complicated, and the instructions are not as clear as they might be (especially for Elm I think).

Maybe we could jump on a call sometime and discuss how to make it clear what the student is supposed to do?

[ceddlyburge]

Hmmm, none of those look that good. Maybe we could create a new exercise. Json encoding / decoding is something that people do find confusing when starting Elm, so it is a good idea. Maybe encoding a request to the github api (which is well known), and decoding a response?