From jamie@zfatal.cern.ch Thu Apr  6 15:19:34 1995
Received: from dxmint.cern.ch by dkuug.dk with SMTP id AA27493
  (5.65c8/IDA-1.4.4j for <sc22wg5@dkuug.dk>); Thu, 6 Apr 1995 13:20:37 +0200
Received: from zfatal.cern.ch by dxmint.cern.ch (5.65/DEC-Ultrix/4.3)
	id AA22420; Thu, 6 Apr 1995 13:19:41 +0200
Received: by zfatal.cern.ch (AIX 3.2/UCB 5.64/4.03)
          id AA38765; Thu, 6 Apr 1995 13:19:35 +0200
From: jamie@zfatal.cern.ch (Jamie Shiers)
Message-Id: <9504061119.AA38765@zfatal.cern.ch>
Subject: Draft Interoperability Report (fwd)
To: sc22wg5@dkuug.dk
Date: Thu, 6 Apr 1995 13:19:34 +0200 (METDST)
Mime-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Content-Length: 1840      
X-Charset: ASCII
X-Char-Esc: 29


I originally sent this to the interop and OOF groups, but received
no comments so far.

I wondered if it might be of interest to a wider audience?

> 
> >                            Interoperability
> >
> > What is interoperability?  Is it Fortran procedures calling C procedures?  Is
> > it making Fortran CORBA/SOM/OLE compliant?  The first is too narrow; the
>  second
> > too broad.
> >
> > CORBA, SOM, OLE are at a higher level of abstraction than any particular
> > language.  A language needs to access the APIs defined by these abstractions,
> > but the abstractions don't have to be incorporated into the language itself.
> >
> 
> Whilst I agree that CORBA, ODMG-93 etc. are at a higher level than
> individual languages, language bindings are, however, defined by
> these "standards".
> 
> The ODMG-93 "standard" for example defines both the C++ and the Smalltalk
> bindings.
> 
> CORBA 2 defines the C++ binding.
> 
> If one believes that object technology and in particular the OMG and
> ODMG "standards" will be of any importance, then one must surely also
> consider the language binding. Of course, it can be argued that this
> is not the job of WG5.
> 
> However, if one wishes to use object request brokers, object databases
> etc. then one will tend to do so from a language that has a defined
> binding.
> 
> Even if we restrict ourselves to something like "the ability to
> call Motif from Fortran" this is not enough. The bulk of the
> Motif code that has drifted across my screen has been generated.
> Therefore, we also need support from these tools.
> 
> Clearly, people who need to use Motif now, or at any time until
> Fortran 2K conforming compilers are widely available, are not
> simply going to wait.
> 
> Playing "catchup" would appear to be an extremely dangerous game.
> 
> --
> Jamie
> 
> 
> 


-- 
Jamie