Allergies "invalid" input test is confusing

[samjonester]

I'm opening this issue for discussion, as I'm not sure what the best solution is, and would like some feedback before opening a PR.

The test case below is confusing. It doesn't make sense why the expected response for an "invalid" input is all by one of the allowed data types. If we're going to expect error handling of this manner, then at a minimum, it should be documented better in the Readme. I'd prefer to just drop this test case, as I don't think it adds value to this exercise, though.

testCase "ignore non allergen score parts" $
    [ Eggs, Shellfish, Strawberries, Tomatoes, Chocolate
    , Pollen, Cats ] @=? allergies 509

[petertseng]

It doesn't make sense why the expected response for an "invalid" input is all by one of the allowed data types.

It is something that I can explain, though the explanation won't necessarily make it make sense:

Prelude Data.Bits> 509 .&. 255
253

So the answer for 509 is the same as what it would be for 253.

If we're going to expect error handling of this manner, then at a minimum, it should be documented better in the Readme.

Agreed. The place to discuss is at exercism/problem-specifications#224

I'd prefer to just drop this test case, as I don't think it adds value to this exercise, though.

In this case, the place to make the PR would be https://github.com/exercism/x-common/blob/master/allergies.json (in addition to this track)

[samjonester]

Thanks for the explanation, @petertseng! That makes sense. It does leads me to believe that the test was written for the specific implementation, which I think is something that is anti-TDD. From that explanation, I'm leaning towards your second suggestion, submitting a PR to allergies.json. Let's see where this conversation leads :)

[rbasso]

It is not closed yet, but exercism/problem-specifications#224 was solved by merging exercism/problem-specifications#389, so I think it is safe to close this issue now that description.md explains the behavior of testing for 509.

Of course, I'll reopen it immediately if there is anything else to be discussed here.