So, you're the kind of guy who downvote the other one who tries to help you? Maaaaaan, that's pretty fair, don't you think? Especially when that guy isn't even the author of the kata or the translation and just take on his time to help you...

FYI, I just checked again, and the solution I used to replace the original one returns your """real convex hull""" as it should. So I don't know what is the problem, but I'm sure of one thing: this time it's on your side. I just changed one thing more, so that the expected value is computed before your function even touches to the data... Just in case you're actually mutating the input without even thinking about it and then reject the fault on someone else...?

Have you reset the kata? The current reference solution returns the second array ("real convex hull" as you call it).

Confirmed. Though, it seems you found a rare edge case, because I check the original solution against one that solves this case correctly and I couldn't make it fail. Well, I replaced the internal solution with the good one, should be good, now.