Orbifx's logarion

Int of float hazard

| | | topics: computing | keywords: ocaml
id: c0615a9d-3844-4b56-91cc-4b1968ddd7ab

Watch out when converting from float to int in OCaml, as there can inconsistent and unintuitive behaviour. OCaml has rarely given me hassle with subtle and obscure issues whilst developing, but this has happened.

I don't remember my case, but I can offer some examples from caml-list.

Example was posted on caml-list by rixed:

int_of_float (float_of_int max_int) = -4611686018427387904

Examples by Nicolas Ratier:

# int_of_float (infinity);;
- : int = 0
# int_of_float (neg_infinity);;
- : int = 0

There was a bug raised about this years ago at https://caml.inria.fr/mantis/view.php?id=4207, but it was not deemed a bug. xleroy explains how this behaviour was inherited by C.