Besides, he should have used const x,y.
So the code is not portable and potentially worse than the fmod function in math and also worse than the current overloaded operators.
And fmod is fmod: floating point modulo.... So it is wrong anyway. Examine the math unit and check what the compiler actually generates for you. You will be surprised.
Note that simple mod works also in FPC (not in delphi) with floats.
Also note that you need integer mod, not fmod...