Date: Wed, 7 Nov 2001 01:38:06 -0500 (EST)
From: Michael Meeks <michael@ximian.com>
To: Owen Taylor <otaylor@redhat.com>
Cc: Gtk Hackers <gtk-devel-list@gnome.org>
Subject: more plug oddness ...


Hi Owen,

        Everything is lovely out of proc with the new socket setup
ie. your suggested:

> gtk_plug_new (0); gtk_socket_add_id (gtk_plug_get_id (plug))?

        My only problem with this scheme, is working out how
plug->socket_window gets set; in the in-proc case, so that when
we hit:

void
_gtk_plug_add_to_socket (GtkPlug   *plug,
                         GtkSocket *socket)
...
  if (GTK_WIDGET_REALIZED (widget))
    gdk_window_reparent (widget->window, plug->socket_window, 0, 0);

        plug->socket_window holds non-NULL and we have a chance of not
just unparenting our GdkWindow.

        This looks a tad daft but FALSE == 0 -> NULL:

-  plug->socket_window = FALSE;
+  plug->socket_window = NULL;

        I'm also having issues with the plug's gdk window not being
shown correctly; I have to have the following code to actualy see an
embedded widget at all with the suggested sequence ( in-proc ):

static void
realize_cb (GtkWidget *socket, gpointer user_data)
{
        GtkWidget *plug, *w;

        g_warning ("Realize");

        plug = gtk_plug_new (0);
        w = gtk_button_new_with_label ("Baa");
        gtk_widget_show_all (w);
        gtk_widget_show (plug);
        gtk_container_add (GTK_CONTAINER (plug), w);
+++     GTK_PLUG (plug)->socket_window = GTK_WIDGET (socket)->window;
        gtk_socket_add_id (GTK_SOCKET (socket),
                           gtk_plug_get_id (GTK_PLUG (plug)));
+++     gdk_window_show (GTK_WIDGET (plug)->window);
}

...
                socket = gtk_socket_new ();
                g_signal_connect (G_OBJECT (socket), "realize",
                                  G_CALLBACK (realize_cb), NULL);
                gtk_widget_show (GTK_WIDGET (socket));
                gtk_box_pack_start (GTK_BOX (vbox), socket, TRUE, TRUE, 
2);
...

        I have no clue why I need to show the gdk window to see ought,
the socket_window kludge is detailed above.

        Am I being dopey again, or are these genine bugs, I'd really
like to clean up bonobo_control_frame_set_remote_window. It seems that
the suggested method is not actualy exercised in gtk+/test/testsocket
which would perhaps be good.

        Regards,

                Michael.