[Caml-list] Re: Interfacing with C: bad practice

[Jeffrey Scofield <jeffsco@psellos.com>]

Gerd Stolpmann <info@gerd-stolpmann.de> writes:

> I don't know for SML, but OCaml also leaves the order unspecified in
> which function arguments are evaluated (and ocamlc and ocamlopt behave
> even differently in this respect). So I think the problem would
> translate to OCaml in some form.

This is a good point.

A problem with C (and almost every other language) is that there's a
lot of room for debate about what the *standard* means (as opposed to
the meaning of particular programs).  With the ML family you have a
formal framework (lambda calculus, I guess) that makes things quite
a bit less ambiguous.  You can still have unspecified parts of the
semantics, but at least it's clearer where the unspecified parts are!

I don't know offhand whether parameter evaluation order is defined
for Standard ML, either.  But there's a very clear specification that
you can look at to find out.  (It seems like the sort of thing that
would be defined in SML.)

Jeffrey