Re: gtk_signal_connect() and passing parameters

[Markus Lausser <sgop users sourceforge net>]

On Wed, May 16, 2001 at 10:22:17AM +1000 Mazur Przemyslaw wrote:

> Hello,
>
> I'm having trouble passing any kind of parameters to
> my call back functions.  For example:
>
> int i = 1456;
> ...
> gtk_signal_connect(GTK_OBJECT(button1),
> "button_press_event",
> GTK_SIGNAL_FUNC(button1_callback), (gpointer) &i);
> ...
> void button1_callback(GtkWidget *widget, gpointer
> data)
> {
>   g_print("i: %d\n", *((int *)data));
> }
>
> Quite simply, I expect to see g_print() print 1456 as
> the value of i.  It does not.  Instead, it prints 4. 
> Even more strange, if I double click on the button
> (which is a checkbox button), it prints 5 instead.
>
> This leads me to believe that what I'm getting via the
> data parameter is something other than the address of
> my i variable.  I tried passing the value of i instead
> (removing the &), and changing to g_print("i: %d\n",
> data); but this gave me another bizzare value instead.
>
> I've had two experienced programmers (although not
> experienced in GTK) look at this and both are stuck
> for a solution.
>
> Anyone have ideas?  I'm starting to think there's some
> bug in my version of GTK.

What you do is: you pass the pointer to a local variable to the callback
function. Thats not a good idea.
instead you have to pass the _value_ itself:
  GTK_SIGNAL_FUNC(button1_callback), (gpointer) i);
and access it with:
  g_print("i: %d\n", (int)data);

Note: if you cast an int to a pointer you should be sure, that 
  sizeof(gpointer) >= sizeof(int)
Btw. i dont know any architecture where this is false, but that doesnt mean,
that there isnt any.