Floating point comparison should use an epsilon

[carldebilly]

When comparing double or single (floating point types), the comparison should be done using an epsilon instead of a straight equality.

https://en.wikipedia.org/wiki/Machine_epsilon