Re: gep-1

[Mark McLoughlin <mark skynet ie>]

Hi Gergo,

On Tue, 27 Aug 2002, ERDI Gergo wrote:

> On 27 Aug 2002, Michael Meeks wrote:
>
> > 	I agree; it's better to have a separate ORBit2-C++ package that works
> > (default build), than to impose complexity on the builder IMHO. I
> > personally, would only really accept C++ bindings in the default build
> > of ORBit2, or outside of it
>
> I still haven't seen any analysis of whether we really have a choice here
> -- is it technically possible to, within ORBit2 proper's API constraints,
> export enough stuff for ORBit2/C++ to be separatable?

	Okay, I had presumed that ORBit-cpp (like the scripting
bindings currently) would be using a lot of internals. It seems not -
I've just committed a patch to make it use CORBA_ORB_create_*_tc
instead of CORBA_TypeCode_allocate (an internal function) and now
ORBit-cpp builds without -DORBIT2_INTERNAL_API.

	So, if I do the idl-compiler module loading we should be able
to split out ORBit-cpp and it won't have to worry about using unstable
ABIs/APIs. So how does that sound ?

Good Luck,
Mark.