It looks horrible to me. I use a lot of FP math, and I'm not going to pick up the habit of sprinkling these dots everywhere. The fact that the language designers would even seriously consider this solution tells me it's not for me.
YMMV, of course.
Definitely not for you, you should look into using C++ if all you do is floating point math routines.
If you use Core, you can use the Float module together with a local open to get nice operators.
open Core
let () =
let n = Float.(2.0 + 3.0) in
printf "%f" n