There should be no floating point repesentation issues. Order of evaluation is defined, so there should be no ambiguity.

It is quite possible that expected values or reference implementations use wrong orders internally, but this problem should be solved correctly. Using fixed point libraries is not the correct solution.

Closing.

Hi. You could post it as a suggestion in the kata Discourse page rather than as a comment to your own solution (it's almost invisible here). I think it's a good idea.