[Caml-list] Re: Legality of using module types from .mli in .ml

[Thorsten Ohl <ohl@physik.uni-wuerzburg.de>]

Julien Signoles <Julien.Signoles@lri.fr> writes:

> The previous example doesn't break my defunctorizer :
> 
> 	  $ ocamldefun --version
> 	  1.02
> 	  $ ocamldefun a.mli a.ml
> 	  $ cat a.ml
> 	  module type T = A.T
> 	  module M = struct let n = 42 end
> 
> That's ok (normal because a.mli is still in the ocamldefun environment
> when a.ml is defunctorized)...

The O'Caml compiler will later not be able to resolve A.T if you use
the `-d' option:

    ohl@wptx47:~misc$ ocamldefun -v
    1.02
    ohl@wptx47:~misc$ cat a.mli
    module type T = sig val n : int end
    module M : T
    ohl@wptx47:~misc$ cat a.ml 
    module type T = A.T
    module M : T = struct let n = 42 end
    ohl@wptx47:~misc$ ocamldefun -d defun a.mli a.ml
    ohl@wptx47:~misc$ cd defun/
    ohl@wptx47:~defun$ ls
    a.cmd  a.ml
    ohl@wptx47:~defun$ cat a.ml
    module type T = A.T
    module M = struct let n = 42 end
    ohl@wptx47:~defun$ ocamlopt a.ml
    File "a.ml", line 1, characters 16-19:
    Unbound module type A.T

I could ignore appendix B of the manual and create a a.cmi:

    ohl@wptx47:~defun$ cp ../a.mli . 
    ohl@wptx47:~defun$ ocamlc a.mli
    ohl@wptx47:~defun$ ocamlopt a.ml

but I'm not supposed to do that, am I?

BTW: given the (currently) uncertain semantics, I have started to
remove the construct from my applications.
-- 
Thorsten Ohl, Physics Dept., Wuerzburg Univ. -- ohl@physik.uni-wuerzburg.de
http://theorie.physik.uni-wuerzburg.de/~ohl/     [<=== PGP public key here]
-------------------
To unsubscribe, mail caml-list-request@inria.fr Archives: http://caml.inria.fr
Bug reports: http://caml.inria.fr/bin/caml-bugs FAQ: http://caml.inria.fr/FAQ/
Beginner's list: http://groups.yahoo.com/group/ocaml_beginners