Re: [orbitcpp-list] Re: cpp branch: CORBA_Object struct hidden?

[Michael Meeks <michael ximian com>]

Hi Richard,

	Having claimed I knew what I was talking about - it seems I'm a totally
wally; so ...

On Thu, 2002-02-21 at 22:37, Richard Andrews wrote:
> CORBA::XXX_ptr 
> CORBA::XXX_ptr::narrow( const CORBA::Object_ptr & rhs )

	You can override a narrow operation such that it equates to a method
invocation ? and the compiler will chain multiple narrows if you narrow
(A -> B -> C) C to A ? fascinating.

	That would indeed localize the internal CORBA_Object, but could a
similar effect not be achieved by having an overrided assignment to
CORBA_Object from Foo:C++Object ? - [ this would be nice also since we
need to do a load of casts to / from CORBA_Typecodes  etc. ].

	So; hmm ... we could then swing this as long as no C++ method that used
data on the inherited instance was inline on the Foo:C++Object. Clearly
this might well have some performance implications, so I think it's best
to go with a non-inherited wrapper - as before ;-)

	Regards,

		Michael.

-- 
 mmeeks@gnu.org  <><, Pseudo Engineer, itinerant idiot