Confused by FFI

[Ian Lynagh]

On Mon, Jul 08, 2002 at 10:19:20PM +0100, Alastair Reid wrote:
> 
> >> For what it's worth, I ran Ian's program on Hugs (with minor
> >> changes: Hugs gets confused by the use of type synonyms in ffi
> >> decls)
> 
> > Is this a hugs bug or am I being non-portable?
> 
> I'd class it as a reflection of the immaturity of Hugs' new ffi
> implementation - but others might reasonably call it a Hugs bug :-)

Sorry if my language was too strong, I just wanted to check I was
working within the FFI spec.

> For them that care, the problems in Hugs are:
[snip]

Thanks.

> > I reached this point having discovered that the Haskell equivalent
> > of "keypad(stdscr, TRUE)" (insert digression/rant about having to
> > assume bool is a numerical type in curses here) had no effect but
> > using the result of initscr worked fine.
> 
> I presume the rant would be to the effect that it's painful that the
> ffi doesn't handle macros such as TRUE or perhaps the plausible but not
> always correct use of 0/1 to marshall False/True?  

It was more directed at the lack of assumptions curses allows you to
make than the Haskell side, but even so the answer is "not really".

> The short answer is that GreenCard and friends exist for that very
> purpose.

I'm currently using hsc2hs and have code like this:

type NBool = #type bool
cTRUE :: NBool
cTRUE = #const TRUE

The problem comes when bool is, say, a 72bit structure and #type can't
provide a suitable Haskell type.

I'm happy to be corrected on this, particularly if it makes my life
easier  :-)

I'll have a better look at GreenCard (I think I decided it was more
complex than hsc2hs for what I was doing at the time, but I might need
the extra complexity now) some time and see if it can help me out.


Thanks
Ian