Feat: Add D&D Character exercise

[khalidbelk]

Hi :) ! Here's a port of the D&D character practice exercise (currently not implemented in the OCaml track). Let me know if there's something missing ; )

[khalidbelk]

@kahgoh could you take a look at this please : ) ?

[kahgoh]

At a glance, this generally looks ok and there are some minor comments below. However, the OCaml track also uses a test generator to generate the tests from the problem specifications. Would you like to try creating one for D&D character?

[khalidbelk]

At a glance, this generally looks ok and there are some minor comments below. However, the OCaml track also uses a test generator to generate the tests from the problem specifications. Would you like to try creating one for D&D character?

The thing is that I initially tried to generate them from a .tpl file but it's a little bit tricky for this exercise, something like the following example will never reach the description fields inside the nested cases array, since the D&D specification structure is kinda inconsistent (uses a nested cases array of objects for the first tests, and then objects that don't follow the same structure), which either causes it to ignore those objects or fails when it doesn't meet an expected field.

...
let tests = [
  {{#cases}}
    {{#cases}}
      "{{description}}" >::
        ae 1 2
    {{/cases}}
  {{/cases}}
]
...

After looking into it a bit more last night I found that it might be possible to preprocess the JSON before the test generator passes it to mustache to avoid those types of errors. Can you confirm that's what it's used for?
I can take a look at it when I have some time :)

[kahgoh]

After looking into it a bit more last night I found that it might be possible to preprocess the JSON before the test generator passes it to mustache to avoid those types of errors. Can you confirm that's what it's used for?

Yes, that's right! You can use the special cases to pre-process the JSON as needed.

[kahgoh]

The changes from your last set of commits look good. Just waiting to see how you go with the test template.

[khalidbelk]

The changes from your last set of commits look good. Just waiting to see how you go with the test template.

Thanks for being active ! I didn't really have time to keep you updated 😅.

I've been investigating further and I think there's no way to just generate all the tests correctly via the generator in the current state of the files, unless:

• 1. The spec JSON is modified (which could potentially break other generators).
• 2. This part of the generator is "modified" to handle these type of complex cases (which obviously isn't a viable option for an issue related to only 1 exercise)

This is because the errors happen in the case-sanitizing step, so before the contents can reach the edit_parameters section.

So, long story short : the issue here is caused by the incompatibility between the inconsistent spec's JSON, and the way the OCaml test generator handles it (not really its fault since it is made to handle consistent JSON's), which prevents those cases from being extracted.

For other exercises, even if they have multiple levels of indentation / multiple cases, they don't have any issues because their spec is consistent. For example see this spec , and its template.

[kahgoh]

Ok, I think we can still go ahead for now. Thanks for looking into the test template!