Re: GEP 5: New Object Adaptor for ORBit2

[Michael Meeks <michael ximian com>]

Hi Mark,

On Thu, 2002-09-05 at 13:16, Mark McLoughlin wrote:
> 	Okay, so I obviously was on crack writing this part. For ABI
> compatibility reasons we obviously can't implement BonoboUnkown as a
> sub-class of ORBitGServant and then magically switch all BonoboObjects
> over to BonoboUnknown. I can't think of a way around this ...

	Well; I havn't had time to look at the code on the branch in any great
detail; however - it seems to me reasonable - for purposes of
implementation inheritance, to have virtual methods that are not exposed
through CORBA. Thus - it would seem necessary to provide some sort of
epv offset into the class member - thus (presumably) if we could arrange
the servant thing to be a BonoboObject itself [ by whatever great feats
], we could make it stick.

	There is the minor [;-)] problem that a bonobo client can have the
servant layout hard-coded into it via the bonobo_object_from_servant
macro - which mostly sucks for any compatibility thing of this sort.
Indeed, short of passing the wrong pointer to the client, this pretty
much scuppers any compatibility between the two, since it assumes that
servant and impl are different entities.

	Hmm,

		Michael.

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