This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
-frepo does not work correctly
- To: egcs-bugs at cygnus dot com
- Subject: -frepo does not work correctly
- From: Kevin VanHorn <kvanhorn at fonix dot com>
- Date: Mon, 26 Oct 1998 17:06:35 -0700
- Organization: Fonix
Version of egcs:
egcs-2.90.27 980315 (egcs-1.0.2 release)
Source:
Attached as file test.tgz (gunzip and untar this to produce the
Libtest directory with Makefile and source files.)
Command arguments given compiler:
See Makefile. The problem occurs when doing "make testmin".
Type of machine and OS:
Intel Pentium Pro
RedHat Linux 5.1
Contents of /proc/version:
Linux version 2.0.34 (root@porky.redhat.com) (gcc version 2.7.2.3)\
#1 Fri May 8 16:05:57 EDT 1998
Operands given to configure command when compiler installed:
Unknown. Installed from RedHat .rpm file.
Description of problem:
When I run "make testmin", I get the following output:
Libtest$ make testmin
g++ -g -frepo -c testmin.cpp
g++ -g -frepo -o testmin testmin.o -lm
collect: recompiling testmin.cpp
collect: relinking
collect: recompiling testmin.cpp
collect: relinking
collect: recompiling testmin.cpp
collect: relinking
collect: recompiling testmin.cpp
collect: relinking
testmin.o: In function `bool
brent<reffct<fun1d<countevals<pointer_to_unary_func
tion<double const *, double> >, double> >,
double>(reffct<fun1d<countevals<pointer_to_unary_function<double const
*, double> >, double> >, double, double, double, double, double, double,
double &, double &, double, double, double, unsigned int)':
/usr/home/kvanhorn/Src/Libtest/min.h:163: undefined reference to
`compute_tol__H1Zd_X01N80_X01'
/usr/home/kvanhorn/Src/Libtest/min.h:169: undefined reference to
`compute_tol__H1Zd_X01N80_X01'
collect2: ld returned 1 exit status
make: *** [testmin] Error 1
But compute_tol is defined in min.h, which is #include'd into
testmin.cpp.
I have found that if I hack the testmin.rpo file, changing the "O"
in front of compute_tol_H1Zd... to a "C", re-make,
then fix the same problem with function lead_quad_coeff by hacking
the .rpo file again, I can get rid of the link errors. But I
shouldn't have to be hacking the .rpo file like this...
test.tgz