(considering the actual task, I guess it's different enough)

The specs about how to redistribute the "non even" values must be given in the task section, not only in the examples.

.

(Î'm pretty sure this is a dupllicate, btw)

isn't the last example of the description false?

When chances are equal and can't be evenly distributed. Follow the original order of chances array

Given [22, 9, 22, 22, 9, 2], 14

It should return [25, 11, 25, 24, 11, 4]
                  +3  +2  +3  +2  +2  +2
                      ^^^^^^

looks like everything is fine

Can you verify if it's ok now. Make sure to reset the kata in the editor.

That's an issue of the kata. EDIT: ok you logged an issue :) EDIT EDIT: and I fixed it ;)

In random tests, the same array is used for both the user's solution and the reference solution, so the user can change the original array, for example using the map! method, which will affect the tests. It is worth using a duplicate array in tests, or write in the description that the user is prohibited from changing the source array

I figured out what the problem was, I used the map! and in tests the same array is used for both the solution and the reference solution

I see, somehow the reference solution passes all tests. So maybe we are missing something. To be honest, I don't even understand the description of this kata. I passed by trial and error.

Yes, I described it too vaguely) In random tests, the added values are doubled

Must be

[[26, 54, 3, 4, 1], 12] => [26+3, 54+3, 3+2, 4+2, 1+2]

random tests require

[[26, 54, 3, 4, 1], 12] => [26+6, 54+6, 3+4, 4+4, 1+4]

I don't understand what you are saying, but it does look random tests are complete BS

The sample tests from the example do not coincide with the logic of random tests, or are insufficient: sample tests assume the result as probabilities totaling 100%, random tests assume probabilities totaling 100% + the dropped probability

[[26, 54, 3, 4, 1], 12]

expected: [32, 60, 7, 8, 5]

got: [29, 57, 5, 6, 3]

resolving