Approved the fork.

judging by the recurring comments the wording seems to be causing a lot of confusion though. here is a fork that tries to improve it

There is no problem in the description, your code needs to validate the input.

True, wasn't really thinking of using a method like this in a huge program or O(n) at all.
If the array was never very long than reallocating wouldn't be all that bad.

Also, no need for 'else' statement after 'return' statement

You just turned a O(n) solution into a O(nlgn) solution :p

int max=Integer.MIN_VALUE,min=Integer.MAX_VALUE;

I believe it's better to assign them both to nums[0].
I'm not sure if it uses less memory but it's more readable and logical.

Maybe you're right.
But as for Java kata, it doesn't worth even 9 kyu :-)

I'm relatively new to C++ katas. I dont have a negative opinion on the point values. it's all good practice in my book

Please be more specific. If you think the description is inadequste, suggest a better one. If you can't pass the tests, post your code, properly marked up and spoilered, post a failing test, input, actual and expected output, and mark it Question. ( Mention the language. You did that right. )

Closing.

Not a kata issue then. Closing.

Is there even anything like java.regex?