I would find a ForeignData extension incredibly helpful. This will be crucial if Haskell ever wants to target out of the ordinary systems.

/jve


On Mon, Dec 1, 2008 at 8:38 PM, John Meacham <john@repetae.net> wrote:
On Mon, Dec 01, 2008 at 04:55:14PM -0800, Evan Laforge wrote:
> On Mon, Dec 1, 2008 at 4:39 PM, Andrea Rossato
> <mailing_list@istitutocolli.org> wrote:
> > Hello,
> >
> > I'm writing the bindings to a C library which uses, in some functions,
> > global variables.
> >
> > To make it clearer, those functions need a global variable to be
> > defined. A C program using my_function, one of the library functions,
> > would look like:
>
> I don't think you can use the FFI to declare symbols for C.  One
> not-so-pretty but effective way to do it is create a stub.c with the
> variables declared along with setting functions, then bind those
> functions like any other.

Yes, it is unfortunate this is the case. my ForeignData proposal was
meant to address this:

http://hackage.haskell.org/trac/haskell-prime/wiki/ForeignData

I am not entirely sure about the proposal as described, but I think
something like it should be done.

       John

--
John Meacham - ⑆repetae.net⑆john⑈
_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe@haskell.org
http://www.haskell.org/mailman/listinfo/haskell-cafe