SilvioCVdeAlmeida <[EMAIL PROTECTED]> writes:
>
>Perl-5.8.4 break the same as 5.8.3 did. But when compiling
>Perl without big ints and long doubles support, the test run
>successfully.
>
>As the error messages points to perl.h:1330:
Can you copy/paste the error message?
(I don't have a build with big ints and long doubles.
so I don't see it.)
>1327 /* eg glibc 2.2 series seems to provide modfl on ppc
>1328 and arm, but has noprototype in <math.h> */
>1329 #ifndef HAS_MODFL_PROTO
>1330 long double modfl(long double, long double*);
>1331 #endif
>
>and /usr/include/bits/mathcalls.h:116:
>__MATHCALL (modf,, (_Mdouble_ __x, _Mdouble_ *__iptr));
>
>then that seems to be a long double implementation bug.
>
>I have glibc-2.3.2-11.9. gcc-3.2.2 and 3.3.3 both goes well.
>
>I've no idea where and how it should be fixed, perhaps
>a simple fix at Configure or hints/linux will do?
Well as the DEFAULT is not to use long double I don't see that
Configure or hints really need fixing, other than perhaps
to warn that embedding in C++ may not work with those options.
>
>
>Silvio
>
>
>SilvioCVdeAlmeida wrote:
>>
>> Now I have perl-5.8.3 and gcc-3.2.2 on Redhat-9.
>> I'll get the new packages, and hope it works here too!
>>
>> Nick Ing-Simmons wrote:
>> >
>> > $ g++ `perl -MExtUtils::Embed -e ccopts -e ldopts` a.cc
>> > No errors at all. (This is perl5.8.4, SuSE 9.1, gcc 3.3.3 )
>> >
>> > SilvioCVdeAlmeida wrote:
>> > >
>> > >Last weekend I've posted a question about embedding Perl in a
>> > >C++ library. The compiling error is easily reproduced by just:
>> > >
>> > >$ g++ `perl -MExtUtils::Embed -e ccopts -e ldopts` a.cc
>> > >
>> > >// a.cc
>> > >#include <EXTERN.h>
>> > >#include <perl.h>
>> > >int main(int a, char** b, char** c) {
>> > > return 0;
>> > >}
>> > >
>> > >That goes right if changed into: gcc aplied to C file a.c with
>> > >the same contents.
>> > >
>> > >I've posted as bug, as adviced (#29966).
>> > >
>> > >The error is about C vs. C++ linkage. I can try to help, if
>> > >someone give me a path to it, as I'm new in perl-xs, and XS
>> > >itself.
>> > >
>> > >
>> > >Thank you.
>> > >
>> > >Silvio
