Re: Activation problems

[Seth Nickell <snickell stanford edu>]

OK, so MonikerContext is not a straight-up BonoboObject but a
BonoboForeignObject. I have no idea what this is... *but* there's
special casing for BONOBO_TYPE_FOREIGN_OBJECT in bonobo-object.c that
prevents calling the function that assigns something to the corba_objref
member variable. Amusingly enough, in bonobo-foreign-object.c we find
the following block of code:

          /* The following type check is deactivated because
           * CORBA_Object_is_a doesn't seem to work correctly,
           * considering that it returns TRUE only if the object is
           * exactly of type Bonobo::Unknown, and FALSE if it is of a
           * derived type, while the CORBA spec says it should return
           * TRUE in both cases.  Perhaps there is a bug in ORBit?.. */
#if 0
        CORBA_exception_init (&ev);
        if (!CORBA_Object_is_a (corba_objref, "IDL:Bonobo/Unknown:1.0",
&ev)) {
                if (ev._major != CORBA_NO_EXCEPTION)
                        g_warning ("CORBA_Object_is_a: %s",
                                   bonobo_exception_get_text (&ev));
                else
                        g_warning ("bonobo_foreign_object_new:
corba_objref"
                                   " doesn't have interface
Bonobo::Unknown");
                CORBA_exception_free (&ev);
                return NULL;
        }
        CORBA_exception_free (&ev);
#endif

-Seth

On Wed, 2003-07-30 at 13:35, Seth Nickell wrote:
> On Wed, 2003-07-30 at 10:31, Michael Meeks wrote:
> > 	For the benefit of anyone reading the archives / with similar problems;
> > the IDL was not correctly inheriting from Bonobo/Unknown - whereas
> > Bonobo/Unknown assumes that in it's vepv construction. Thus the vpev
> > pointed at one thing, and the code at another; such that the vepvcache
> > wasn't built correctly for that instance, and whacked out things
> > happened in-proc when invoking a method.
> > 
> > 	I've added a test for this to libbonobo, unconditionally enabled with
> > it's configured with --enable-debug=yes.
> > 
> > 	An easy mistake to make; next time it'll be an easier one to catch I
> > hope :-)
> 
> I *knew* I'd done something wrong, its one of the things I do best. :-)
> 
> #ifdef G_ENABLE_DEBUG
> 	if (!CORBA_Object_is_a (BONOBO_OBJREF (object),
> 				"IDL:Bonobo/Unknown:1.0", NULL))
> 		g_error ("Attempt to instantiate non-Bonobo/Unknown "
> 			 "derived object via. BonoboObject");
> #endif
> 
> So right now this causes failures in programs that initialize Bonobo,
> which is a bad thing... The problem is (as you said) creating the
> MonikerContext object. It looks like what is happening is that
> object->corba_objref is NULL, which is of course causing
> CORBA_Object_is_a to harf when BONOBO_OBJREF consequently returns NULL.
> Is this happening too soon (i.e. before all the corba stuff is connected
> together) or is something unusual about the monikercontext object? I
> don't know enough about Bonobo internals etc to really be able to push
> farther I think.
> 
> -Seth
> 
> _______________________________________________
> gnome-components-list mailing list
> gnome-components-list gnome org
> http://mail.gnome.org/mailman/listinfo/gnome-components-list