Package: g++-3.2
Severity: important
Version: 3.2.1-0pre5
As James has already posted, the following does not work on sparc:
| (unstable)troup@vore:~$ cat test.C
| #include <cstdio>
|
| int main (int argc, char **argv)
| {
| printf("Hello\n");
| return 0;
|
| }
| (unstable)troup@vore:~$ echo "#include <cmath>" > test-with-cmath.C; cat test.C >> test-with-cmath.C
| (unstable)troup@vore:~$ g++ -o test{,.C}; echo $?
| 0
| (unstable)troup@vore:~$ g++-3.2 -o test{,.C}; echo $?
| 0
| (unstable)troup@vore:~$ g++ -o test-with-cmath{,.C}; echo $?
| 0
| (unstable)troup@vore:~$ g++-3.2 -o test-with-cmath{,.C}; echo $?
| In file included from test-with-cmath.C:1:
| /usr/include/c++/3.2/cmath: In function `long double std::acos(long double)':
| /usr/include/c++/3.2/cmath:109: `::acosl' undeclared (first use here)
| /usr/include/c++/3.2/cmath: In function `long double std::asin(long double)':
| /usr/include/c++/3.2/cmath:127: `::asinl' undeclared (first use here)
| /usr/include/c++/3.2/cmath: In function `long double std::atan(long double)':
| /usr/include/c++/3.2/cmath:145: `::atanl' undeclared (first use here)
| /usr/include/c++/3.2/cmath: In function `long double std::atan2(long double,
| long double)':
| /usr/include/c++/3.2/cmath:164: `::atan2l' undeclared (first use here)
| /usr/include/c++/3.2/cmath: In function `long double std::ceil(long double)':
| /usr/include/c++/3.2/cmath:183: `::ceill' undeclared (first use here)
| /usr/include/c++/3.2/cmath: In function `long double std::cosh(long double)':
| /usr/include/c++/3.2/cmath:211: `::coshl' undeclared (first use here)
| /usr/include/c++/3.2/cmath: In function `long double std::exp(long double)':
| /usr/include/c++/3.2/cmath:229: `::expl' undeclared (first use here)
| /usr/include/c++/3.2/cmath: In function `long double std::floor(long double)':
| /usr/include/c++/3.2/cmath:257: `::floorl' undeclared (first use here)
| /usr/include/c++/3.2/cmath: In function `long double std::fmod(long double,
| long double)':
| /usr/include/c++/3.2/cmath:276: `::fmodl' undeclared (first use here)
| /usr/include/c++/3.2/cmath: In function `long double std::frexp(long double,
| int*)':
| /usr/include/c++/3.2/cmath:295: `::frexpl' undeclared (first use here)
| /usr/include/c++/3.2/cmath: In function `long double std::ldexp(long double,
| int)':
| /usr/include/c++/3.2/cmath:315: `::ldexpl' undeclared (first use here)
| /usr/include/c++/3.2/cmath: In function `long double std::log(long double)':
| /usr/include/c++/3.2/cmath:334: `::logl' undeclared (first use here)
| /usr/include/c++/3.2/cmath: In function `long double std::log10(long double)':
| /usr/include/c++/3.2/cmath:352: `::log10l' undeclared (first use here)
| /usr/include/c++/3.2/cmath: In function `long double std::modf(long double,
| long double*)':
| /usr/include/c++/3.2/cmath:376: `::modfl' undeclared (first use here)
| /usr/include/c++/3.2/cmath: In function `long double std::pow(long double, long
| double)':
| /usr/include/c++/3.2/cmath:410: `::powl' undeclared (first use here)
| /usr/include/c++/3.2/cmath: In function `long double std::sinh(long double)':
| /usr/include/c++/3.2/cmath:451: `::sinhl' undeclared (first use here)
| /usr/include/c++/3.2/cmath: In function `long double std::tan(long double)':
| /usr/include/c++/3.2/cmath:479: `::tanl' undeclared (first use here)
| /usr/include/c++/3.2/cmath: In function `long double std::tanh(long double)':
| /usr/include/c++/3.2/cmath:497: `::tanhl' undeclared (first use here)
| 1
| (unstable)troup@vore:~$
The problem is caused by
/usr/lib/gcc-lib/sparc-linux/3.2.1/include/bits/mathdef.h:77:
#ifndef __NO_LONG_DOUBLE_MATH
# if __WORDSIZE == 32
/* Signal that in 32bit ABI we do not really have a `long double'.
The disables the declaration of all the `long double' function
variants. */
# define __NO_LONG_DOUBLE_MATH 1
# endif
when combined with:/usr/include/c++/3.2/sparc-linux/bits/c++config.h:398:
/* Define if you have the acosl function. */
#define _GLIBCPP_HAVE_ACOSL 1
(and other #defines like it that cause the above errors)
So, all of the functions that are causing an error need to be wrapped in
# ifndef __NO_LONG_DOUBLE_MATH - 0 blocks in c++config.h, or in
# __WORDSIZE checks. This may mean that sparc needs an arch specific
# c++config.h, like it has an arch specific mathdef.h
However, all of these functions do exist in libm.so.6, so perhaps the sparc
specific bits/mathdef.h should be removed instead?
Ben? What should we do with this?
--
Ryan Murray, Debian Developer (rmurray@cyberhqz.com, rmurray@debian.org)
The opinions expressed here are my own.
Attachment:
pgpiLdTyepeqR.pgp
Description: PGP signature