Original bug ID: 4149
Reporter: schneck
Status: closed (set by @damiendoligez on 2006-11-15T11:19:43Z)
Resolution: won't fix
Priority: normal
Severity: minor
Version: 3.09.2
Category: ~DO NOT USE (was: OCaml general)
Duplicate of: #4025

Bug description

We have found that on an Intel Mac, we will get different behavior with some floating point operations in native code and bytecode. Here is a minimal example:

(*
On Intel Mac compiled in native code:
4600877379321698713
4600877379321698714
but in bytecode:
4600877379321698714
4600877379321698714
*)

let print f = print_endline (Int64.to_string (Int64.bits_of_float f))

let main () =
let x = 1. /. 3. in
let y = 1. /. 2. in
print (x /. (x +. y));
let x_plus_y = x +. y in
print (x /. x_plus_y)

let _ = main ()