Re: [4.0.0] "make check-g++" fails linking tests on powerpc-apple-darwin7.9.0

[Andrew Pinski <pinskia at physics dot uc dot edu>]

On Jun 21, 2005, at 11:53 PM, Douglas Gregor wrote:

> Running "make check-g++" for GCC 4.0.0 fails any tests that need to  
> link on powerpc-apple-darwin7.9.0, with error messages such as:
>
> /usr/bin/ld: warning suggest use of -bind_at_load, as lazy binding may  
> result in
>  errors or different symbols being used
> symbol _tanl used from dynamic library  
> /Volumes/scratch/dgregor/gcc-concepts-con
> fig/powerpc-apple-darwin7.9.0/./libstdc++-v3/src/.libs/ 
> libstdc++.dylib(single mo
> dule) not from earlier dynamic library  
> /usr/lib/libSystem.B.dylib(ld64.o)
>
> We've seen these before on powerpc-apple-darwin7.x.0 (x < 9), but they  
> were always fixed by updating cctools to something > cctools-495. I'm  
> using cctools-576. The same problem occurs with mainline GCC (4.1.0).
>
> Does anyone know how this can be fixed?

I did not see this before 10.3.9.  The problem is that math.h was not  
updated
for the library changes which went into 10.3.9.  There are three ways  
around this,
either downgrade to 10.3.8, fix math.h somehow, or upgrade to Tiger.  I  
have taken
the first approach, well I never updated in the first place.

This is not a bug in GCC or libstdc++ but how Apple handled the update  
which in my
mind was wrong.

Thanks,
Andrew Pinski