Re: Gcc 3.0 Open Issues

[Stan Shebs <shebs at apple dot com>]

Helge Hess wrote:
> 
> "Bryan W. Headley" wrote:
> > 4. What's the ObjC demand look like with Mac OS/X and GNUStep
> > clientele? I'm wondering what ObjC++ is, and whether insertion of it's
> > features would take care of the kruft problems that ObjC has.
> 
> As far as I have seen ObjC is still considered part of gcc, which is
> good. The basic features are stable and probably don't need a lot of
> further work. There are some smaller things which would be nice in the
> compiler, but there are no people doing such work, so we are basically
> stuck.
> We'll see how much input comes from Apple here.

There are some things that could be contributed, but they won't be of
much interest until GCC gets a Mach-O / OS X backend.  I'm working on
submittable patches for this, but I can't guarantee that they will make
it in soon, so they shouldn't be one of the release criteria.
 
> Regarding ObjC++, I guess this probably won't go into gcc 3.0 since it's
> addition is a significant amount of work and doesn't have a lot of
> support. Apple's ObjC++ is AFAIK based on gcc 2.7.x which makes adding
> the stuff to gcc 3.0 quite difficult. Besides that it doesn't fix any of
> the ObjC problems we have (which are mostly related to dynamic function
> interception and dispatch). ObjC++ allows integration of ObjC and C++
> which is nice but not incredible important (we survived without it quiet
> some time ;-)

Yes, ObjC++ is implausible for addition to GCC, not least because there
is no actual definition of the language, which makes testing difficult... :-)

We're looking at alternative strategies for people with C++ code that
they need to be callable from ObjC and vice versa.

Stan