Re: a little further on Darwin

[David Ayers]

Yves de Champlain wrote:
> > A long time ago ProjectCenter used to treat all compiler warnings as 
> > errors, I changed this last September. If your ProjectCenter code is 
> > older than this, you may just update it.
> > To make this very clear: The -Werror flag is a valueable tool during 
> > development, but should never be put in released makefiles, as you 
> > never can tell which warnings a different version of the compiler may 
> > give.
> >
> > Fred
>
>
> mmm, I missed that one, I have 0.3.5 (07-10-03 in ChangeLog) but this is 
> the latest
> "advertised" version I could find ... should I go for CVS ?  I would 
> rather have
> a release because I'm planning on including a GNUstep bundle (not 
> playing with words)
> in the next version of the GNU-Darwin distribution (I hope to have RC1 
> online by next week).

To be honest, I'm not sure how reasonable this is.  Those undefined 
symbols you are getting are a real issue that are probably show stoppers 
if not resolved.  Do the -base tools (defaults/gdomap/autogsdoc) work? 
Also, are you going to package a static FSF libobjc?  Where are the 
headers installed wrt to Apple runtime headers?

Andrew Pinski is currently trying to sort out the parallel installation 
of the Apple and GNU runtimes on Darwin.  Chances are that with any 
current versions of GCC that you have issues wrt to finding the correct 
headers, libraries to link against, libraries to load at runtime,

There may also be certain issues wrt to shared libraries vs. frameworks 
as the Darwin linker tries to out-smart our simulated frameworks leading 
to issues at runtime.

I'd offer to help you through compiling the core packages correctly if I 
had ssh access to a GNU-Darwin box.  (I don't have any hardware 
available here to install it.)  But I only have very limited time and we 
would have to coordinate this privately.  But I couldn't hold a deadline 
of next week.

Cheers,
David