RE: Java/Bonobo and inter-orb communication [was: shipping Vera w ith 2.4]

[Murray Cumming Comneon com]

> From: ERDI Gergo [mailto:cactus telnet hu] 
> On Tue, 4 Mar 2003 Murray Cumming Comneon com wrote:
> > I don't think we really have a problems with cross-language 
> > implementation inheritance. Bonobomm works already. Our 
> problems are 
> > either
> > a) orbitcpp is difficult to complete.
> > b) replacing orbitcpp with another ORB (e.g. mico) that has 
> C++ already
> > seems difficult.
> 
> c) The CORBA/C++ mapping is far worse (relative to C++'s 
> possibilities) than CORBA/C  :)
> 
> but b) is, when you look at it from another perspective, a 
> perfect example of cross-language implementation inheritance. 
> BTW, I don't really see how we could use another ORB and 
> still re-use libbonobo implementations, unless there was a 
> crazy scheme involving the CORBA-ification of every minute 
> detail of the libbonobo implementation. (Just imagine what 
> kind of IDL's you'd need if you wanted to subclass a
> C++ class with only using CORBA as the communication between 
> the child 
> C++ and
> the parent class)

Well, I think we already do this in bonobomm, regardless of any theoretical
efficiency considerations, but we are of course limited to accessing the
"public" API in the IDL. According to our own C++-ist
implementation-hiding-principles we maybe shouldn't be messing with the rest
anyway.

Murray Cumming
murrayc usa net
www.murrayc.com