okay I forked the kata and I think I fixed it but idk how to get the fix published

Might not be perfect but the sooner we can fix it the better

Yes, this! Especially the second one. I spent hours attempting this kata before realizing the reference implementation is the one that is wrong, not my code.

As long as the test remains broken the way they currently are, the quickfix for passing the broken tests would be to throw out all negative numbers that are also odd.

I'm guessing the mistake made was thinking that i % 2 == 1 holds for all odd numbers; actually: a negative odd number % 2 == -1

Surprisingly fun kata, very enjoyed! The empty array in an array edge case input really confuses me. I was dissapointed when I got the whole thing working except for the empty array in an array test. I guess I don't really understand how to work with an array with only an empty array in it in Java.

I really liked this kata. Some people didn't like the "hardcoding", but maybe some problems genuinely require it and this is a really practical example. Here in Java we've got it down to not much more than a single switch statement (with 30 cases).