Re: PROPOSAL: Objective-C++

[Ovidiu Predescu]

On Thu, 15 Nov 2001 09:07:32 -0800, Stan Shebs <shebs@apple.com> wrote:

> Markus Grabert wrote:
> > 
> > - ANSI Objective-C, with a full definition of the runtime; and
> >   therefore one standard way to combine ObjC with other languages.
> 
> Creating an ANSI standard is a lengthy and expensive process, and
> nobody who cares about ObjC has deep enough pockets to make it
> happen.  That doesn't preclude having a formal quasi-standard,
> and if I weren't helping get GCC 3 to where it can be the default
> compiler for OS X, that would probably be one of the things I'd be
> working on.

I agree with you, Stan, there is no point in going through an ANSI
standardization process. Just look at what happened to Common Lisp,
and how much effort and time it took to standardize it. By the time
they finished it, Common Lisp was not even in the radar screen of many
companies :-(.

IMO when we talk about standardizing Objective-C, we talk about the
runtime system and the language itself. Fortunately the language is
pretty much similar across compilers, as the most widely used compiler
is the GNU compiler. It may be a problem with the POC compiler of
David Stes, we need to talk to him about that.

In terms of the runtime system, I think we need to come up with a set
of functionalities for it. Based on this, we should invent a new set
of functions and classes that could be implemented on top of the
existing runtime systems, that implement the required functionality.

New systems will then be built using the new runtime API, while old
systems can still continue to use the old APIs. One key thing for this
to succeed though, is to have the ability to take the best features
out of the NeXT/Apple runtime and incorporate them in the GNU runtime,
and vice-versa. So from this point of view it won't hurt for Apple to
donate their runtime library to FSF, even if it ends up not being used
as it is in the current implementation. It will at least allow both
the GNU and Apple implementations to freely use the best things out of
the other implementation.

Of course, as Stan pointed out, the new API for the runtime should
have similar licensing APIs as libobjc today. This means that the
programs which link libobjc should not be required to be GPL, or even
LGPL.

Perhaps we should also invite David Stes in the discussion. He might
bring an interesting perspective with his POC work.


Best regards,
-- 
Ovidiu Predescu <ovidiu@cup.hp.com>
http://orion.rgv.hp.com/ (inside HP's firewall only)
http://sourceforge.net/users/ovidiu/ (my SourceForge page)
http://www.geocities.com/SiliconValley/Monitor/7464/ (GNU, Emacs, other stuff)