Just published another fork with 100 random tests. Feels like they would flag similar stuff as the non-random ones anyway.

I actually did that first, but you'd still have to limit the maximum value to some number. Otherwise you might get execution timeouts.

It's best to use 100 random tests.

Also added a test with a very big number.

Thanks Steffan153. The first comment wasn't clear at all - it makes sense now.

Regarding the 1 random test I would completely agree in the real world! But I'm also a fan of consistency and diverging from the original translation wouldn't help anyone who wants to learn how a piece of code gets translated 1:1 from another language (test included). Changing the test makes this not a "translation" anymore. Probably the original (Java) shouldn't be allowed in the first place if that's a problem. Anyway, here is a modified version, which will test all values until 2000: https://www.codewars.com/kumite/5d4c6f84164480002b4581c8?sel=5d4c86439dbe5f001d26d1d8. I played around with the value, but going too high just makes the test time out.

And 1 random test in another translation isn't a good practice to follow...

Gotcha (I think) https://www.codewars.com/kumite/5c3f6d5b460e9b408e780cfd?sel=5d4c6f84164480002b4581c8

It's public, it can be called from solutions. With more than 1 translation its name will be predictable. (If someone really wants to cheat, private isn't the panacea, but it's better than nothing.)

Not sure what the first sentence means - isn't it private?
Regarding tests - I literally translated the original Java kata, including the tests, making it match completely.

The reference solution be private.
How about more than 1 random test?

Approve, please?