Alastair Reid wrote:
> Hmmm, in Hugs I decided to reject any unsupported calling convention
> on the assumption that silently using the wrong one was worse than
> loudly objecting.

I'd prefer that behaviour for GHC, too. The FFI spec doesn't say
anything about that, perhaps it should...

> [...] putting aside, if you can, the fact that you'd have to change
> all your code again if things were 'fixed'?

I don't have to change my code, I'm using the following stuff now in
my Makefile (basically stolen from the network package):

   ifeq "$(TARGETPLATFORM)" "i386-unknown-mingw32"
   SRC_HC_OPTS += -DCALLCONV=stdcall
   else
   SRC_HC_OPTS += -DCALLCONV=ccall
   endif

Additionally one has to remember which foreign stuff comes from a DLL
on WinDoze (using CALLCONV then instead of ccall) and which doesn't.
:-P   Not nice, but I can't see a way around this CPP-trickery.

Cheers,
   S.
_______________________________________________
Cvs-ghc mailing list
[EMAIL PROTECTED]
http://www.haskell.org/mailman/listinfo/cvs-ghc

Reply via email to