[wyldfire]

> a floating point operation raises the "inaccurate" flag, how do you deal with that?

You can trap. ...but then again, existing arithmetic traps are not uniformly enabled by default.