I think you should make it a new kata. Modifying my solution so that it would print out the pieces would be non-trivial.

Heh, our solutions were both counting sort (mine was only so rigorously microoptimised it may not have been transparent), and it pulls 10049 ms in comparison to your 11000, so it wasn't a problem with my code, but rather your tests being too difficult.

This is definitely not 6 Kyu. It should be 5 at the very least.

Well code is first for humans though? As written it's very hard to understand, so I can't even read it well enough to be impressed.

This is very unpythonic code.

Thanks for making those kata, they were quite fun.

This Kata is too low leveled. Using a heap, quickselect, etc seem to fail. I don't expect people without decent algorithmic competence to pass this without massive effort expenditure due to the size of the test cases.