Message-ID: <4296897F.7060007 at rftp.com>
Date: Thu, 26 May 2005 19:44:15 -0700
From: Robert Roessler <robertr at rftp.com>
Organization: Robert's High-performance Software
MIME-Version: 1.0
To: lablgtk at math.nagoya-u.ac.jp
Subject: Additional export request from dlllablgtk2
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit

To "make things work", I implemented the following code in my 
OCaml-C-land widget interface:

#ifdef CAML_DLL
//	create a Val_GtkWidget_sink here since we can't get at the one in 
"ml_gtk.c"
#define gtk_object_ref_and_sink(w) (g_object_ref(w), gtk_object_sink(w))
Make_Val_final_pointer_ext(GtkObject, _sink , gtk_object_ref_and_sink, 
g_object_unref, 20)

//	(supports the Val_GtkWidget_sink impl)
static void ml_raise_null_pointer()
{
	static value * exn = NULL;
	if (exn == NULL)
		exn = caml_named_value ("null_pointer");
	raise_constant (*exn);
}
#endif	/* CAML_DLL */

1) figuring out that I needed the CAML_DLL conditional compilation was 
"expensive" - including this code in a non-DLL ("lib") build caused 
major trouble :(

2) this lets me build and run LablGTK OCaml app, without *apparent* 
trouble - but I am not at all sure that it was the "right" thing to do

3) if Val_GtkWidget_sink was exported from the dlllablgtk2 DLL, then 
apps like mine (which wrap their own widgets) could have access to the 
"real" Val_GtkWidget_sink implementation just like the native builds

4) then I could remove this code entirely! :)

Or do I have this all wrong?

Robert Roessler
robertr@rftp.com
http://www.rftp.com