Comment by TonyTrapp

As funny as it sounds, there is still a bit of truth in there, though: code might change from using floats to double for example, so you might want to replace single-precision constants by double-precision constants. Only need to replace a single pi constant in that case. :)