Re: DO segfault

[Richard Frith-Macdonald]

On Friday, August 15, 2003, at 01:36 PM, Richard Frith-Macdonald wrote:

> On Friday, August 15, 2003, at 12:27 PM, Richard Frith-Macdonald wrote:
>
> > On Friday, August 15, 2003, at 11:58 AM, Stefan Böhringer wrote:
> >
> > > I'm using the following:
> > >
> > > [pingu@hgX mffinder]$ gcc3 -v
> > > Reading specs from /usr/lib/gcc-lib/i686-pc-linux-gnu/3.2.1/specs
> > > Configured with: ../gcc-3.2.1/configure --prefix=/usr
> > > --enable-languages=c,c++,objc,f77 --enable-shared 
> > > --enable-threads=posix
> > > --enable-nls --with-gnu-as --with-gnu-ld --without-included-gettext
> > > --with-system-zlib --enable-version-specific-runtime-libs
> > > --program-suffix=3
> > > Thread model: posix
> > > gcc version 3.2.1
> >
> > I'm using 3.4 ... but I wouldn't call 3.2 old.
> > I have an old (slow) machine with 3.0 on it, so I'm going to update 
> > it to the current GNUstep code from cvs, and try building and running 
> > your zillion test there.
>
> Ok ... I had to hack the zillion code around a bit to get it to 
> compile with gcc-3.0 (that compiler doesn't support forward 
> declaration of protocols), but when I ran the resulting code it 
> crashed.  Looks like this is the old compiler bug of Protocol objects 
> not being properly initialized by the compiler/runtime.  I had thought 
> that this was fixed by gcc-3.2 but it looks like that is not the case 
> :-(  Probably you need 3.3 or later.
>
> I think it is possible to code an (ugly) hack to work around the 
> problem into the NSDistantObjects code.  I may do that after lunch.

OK ... I've done that and comitted it to cvs (actually, a change to 
NSObject.m and one to NSDitantObject.m) after trying it out on my 
gcc-3.0 system.  Please let me know if it works ok for you.