Re: gtk-2 bug, inconsistency or simply misunderstanding on my side ?

[Olexiy Avramchenko <ath beast stu cn ua>]

Olivier Fourdan wrote:

> (resent 'cuz original message was sent from the wrong mailbox 
> - Sorry, list moderators....)
> --
> Hi Owen, Olexiy,
>
> Here comes my example. It's based on gtkdial.
>
> Get the source from http://www.xfce.org/archive/gtkdial.bug.tar.gz
>
> Untar, compile with "make" and run it.
>
> When you click on the button, you get the following message :
>
> ** Message: freeing dial widget (80724f8)
> ** Message: freeing dial widget (80724f8)
>
> (dial_test:5926): Gtk-WARNING **: A floating object was finalized. This
> means that someone
> called g_object_unref() on an object that had only a floating
> reference; the initial floating reference is not owned by anyone
> and must be removed with gtk_object_sink() after a normal
> reference is obtained with g_object_ref().
>
>
> The trace (g_message) is located right in gtk_dial_destroy(). It shows
> that the method is called twice on the same object (80724f8)
>
> PS: The frame widget is destroyed with "gtk_widget_destroy()"
Hello, Olivier
GtkDial is buggy: there's a check before unreffing adjustment in destroy 
handler
if (dial->adjustment)
   g_object_unref(dial->adjustment);
but there's not dial->adjustment = NULL after unreffing, so right one 
code is:
if (dial->adjustment) {
   g_object_unref(dial->adjustment);
   dial->adjustment = NULL;    /* missed there */
}
So, its doesn't matter whether you destroys frame, containing GtkDial or 
GtkDial
tself.
As Owen said there's a possibility that "destroy" handler can be called 
twice,
this happens only to 'real' handlers (pointed by GtkObjectClass->destroy) in
widgets realizations.

Olexiy