Re: C++ oaf patch.

[Maciej Stachowiak <mjs eazel com>]

Michael Meeks <michael@helixcode.com> writes:

> Hi Maciej,
> 
> On 10 Jul 2000, Maciej Stachowiak wrote:
> > Michael Meeks <michael@helixcode.com> writes:
> > >     Just committed this to help our C++ friends =)
> >
> > This is kind of interesting. If you use liboaf from C++, you still get
> > CORBA_Object's, not CORBA::Object's. Is there a way to convert between
> > them?
> 
> 	Not as far as I am aware; no.
> 
> > Otherwise, this is not so great for C++ hackers, and we
> > shouldn't pretend to work for C++ if we don't.
> 
> 	Um; but wait, we still work. The Gnome-libs headers can be used
> from C++ since they do this, true the objects are not named Gnome::Canvas
> but I imagine this doesn't cause too much lost sleep.
> 
> 	I think there is a world of difference between having a cumbersome
> object naming scheme and having to stick ugly extern's in your C++ code to
> include headers that should know better =)

I disagree they should know better. I don't think C++ code should use
the CORBA C language binding. Apparently, ORBit agrees since none of
the public ORBit headers that I can see have extern "C" guards. In
fact, it's kind of bizzare to put these guards on the OAF headers but
not the ORBit headers.

> 	Anyhow; it seems like what you are asking for is a) nice ORBit C++
> wrappers, b) C++ interfaces for oaf which is a far taller order.

There are ORBit C++ bindings, though I personally have no clue how
good they are. A more useful solution for C++ programmers would be an
OAF interface that works with this. This is actually somewhat
problematic because liboaf provides some functionality beyond just
wrapping the CORBA interfaces to oafd; we should look into minimizing
that.

 - Maciej