You misunderstood the behaviour of wrapping (which is, arguably, not described precisely). It is not that when going from positive to negative you always end up with 255. You need to wrap through 255 to lower values, if necessary, so -1 becomes 255, -2 becomes 254, and -12 becomes 244.

L Q Q c U B N
0 -> 2 -> 4 -> 16 -> (copy = 16) -> 4 -> -12 (=255) -> output 255
expected:<2[44]> but was:<2[55]>
Where did I make a mistake?

Haskell translation

I've made an attempt at generating random valid inputs. Loops aren't generated due to the complexity of ensuring that they don't loop forever. However, they should serve the purpose of weeding out any hard-coded solutions.

python new test framework + random tests are required. updated in this fork

Quite funny kata.

Added in latest fork

python new test framework is required. updated in this fork

Nice!

Could be excellent, but.. returning a const char* is bad c++ practice :-(

Re-raised as an issue.

approved at some point

tests are not very torough in JavaScript either, I had to add a fixed test to invalidate my solution. It needs more test cases like [857, 85]

c++ version should use std::string instead of const char*

splendid kata !!!