No, please remove all stuff specific to this Java translation (especially null and negative inputs) and make this translation conform to the other languages.

C++ Translation

Yes, I've been careful to put it in a Java-only block so as not to alter the specification of the problem

I will specify it at the description

Before you change the description, and, in consequence, specification of the problem, make sure that the existing translations agree on the new reqs.

Hi! Thanks for the feedback.

Well, It's true that it doesn't mention anything, because in the initial kata it didn't specify that case, except in the last example. But that input is indeed considered and valid.

I will specify it at the description and also I can instantiate a single static Random for the performance problem.

• I cannot see anything in specs about k > len(nums). Is this a valid input for this problem?
• the new Random() called for every call of rand is a performance problem. It should use either the ThreadLocalRandom instance, or an instance field of the test class, and not recreate it on every call.

Java translation