the kata needs clarification but I think if it meant to return "bye" and not "e" then it should also return "hi" if "hi" is the only argument. otherwise it's inconsistent. what if the arguments were ("hello", [1, "world", [2, 3]])? Should it return [1, "world", [2, 3]], [2, 3], or 3 ?

A zero-liner, anyone?

Modified, thanks!

Have you tried again? I changed the tests.

"Sample tests" in Ruby still have the same problem

Broken how exactly? I passed the kata without problems.

P.S.: I see there are already open issues about that below.

The tests only use a precision of 1e-6 when comparing the results

In all translations results were and are tested with a tolerance of 1.0e-6...
I remove the "you can truncate..." which was a remnant of the first kata.

Sorry: the "Sample tests" have not been updated.
Mpdified. Thanks!

beggars([1,2,3],0) -> [NaN: NaN]. Does not work, should return []. Current tests are testing for n=0 too

Same

No, it does, because Array(0) create empty array and the cycle doesn't begin

;) Take a look at my one-liner.

it's O(n*log(k)), iirc. n the length of the argument and k the number of elements you keep. Comparing to the other approach, it makes it O(n), k being constant. But the overhead of the constant term is far from small (especially because that tool is built in pyhton code, not like the sort functions that are in Cython), so you'll see the effect if k is small enough, and the list long enough.