I would put it this way:
- If you writing code, which requires comparing two floating-point values for equality, then you do something wrong.