135 GCC regressions, 132 new, with your patch on 2003-07-01T05:03:50Z.

Andrew Pinski pinskia@physics.uc.edu
Tue Jul 1 11:52:00 GMT 2003


On Tuesday, Jul 1, 2003, at 07:26 US/Eastern, Andreas Jaeger wrote:
> Jerry, this looks like a problem introduced with your change (see
> below).
>
> Andrew Pinski <pinskia@physics.uc.edu> writes:
>
>> [...]
>> It is not the ISO C conversions that is causing this.
>> The problem is that there are two duplicate symbols in two different
>> *.o in libstdc++.a:
>> ranlib: file:
>> /Users/pinskia/fsf-clean-nocheck/lib/libstdc++.a(concept-inst.o)
>> defines symbol: __ZNSt8numpunctIcE2idE
>> ranlib: file:
>> /Users/pinskia/fsf-clean-nocheck/lib/libstdc++.a(locale-inst.o)
>> defines symbol: __ZNSt8numpunctIcE2idE
>>
>> I cannot pin point the patch which caused this but it looks like a
>> libstdc++ one, rather than a gcc one.
>
> Thanks Andrew.
>
> Looking at the ChangeLog and your results I suggest this one might be
> the culprit:
>
> +2003-06-27  Jerry Quinn  <jlquinn@optonline.net>
> +
> +	* src/locale.cc (__use_cache<numpunct>): Move from here ...
> +	* include/bits/locale_facets.tcc (__use_cache<numpunct>): To
> +	here.
>
> I'm CC'ing the libstc++ mailing list for further discussions,
> Andreas


This patch has already be reverted:
2003-06-30  Jerry Quinn  <jlquinn@optonline.net>

         * src/locale.cc (__use_cache<numpunct>): Revert previous 
relocation.
         * include/bits/locale_facets.tcc (__use_cache<numpunct>): Ditto.

Thanks,
Andrew Pinski



More information about the Libstdc++ mailing list