This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
A clue for the libstdc++ problem.
- To: gcc at gcc dot gnu dot org
- Subject: A clue for the libstdc++ problem.
- From: "H . J . Lu" <hjl at lucon dot org>
- Date: Sun, 1 Apr 2001 10:44:54 -0700
- Cc: aoliva at redhat dot com
If you take a look at glibc, we have to do some extra work to make sure
shared libc.so is built right. We cannot use the same method for
building generic shared libraries. You have to do the same for
libstdc++. I don't think there is an easy solution.
Here is a kludge. It is not a correct solution for the libstdc++
problem. But I hope you can get some clues. I will spend some more
time on it today.
H.J.
----
2001-04-01 H.J. Lu (hjl@gnu.org)
* ltcf-cxx.sh: Make sure we never use g++ for the shared
libstdc++-v3.
--- gcc/ltcf-cxx.sh.libstdc++ Sat Mar 24 20:23:06 2001
+++ gcc/ltcf-cxx.sh Sun Apr 1 10:35:07 2001
@@ -46,7 +46,20 @@ lt_simple_link_test_code='int main(int,
CXX=${CXX-c++}
# ltmain only uses $CC for tagged configurations so make sure $CC is set.
-CC=${CC-"$CXX"}
+#
+# Special case for libstdc++.
+case `pwd` in
+*/libstdc++-v3)
+ if test "x$LTCC" != "x"; then
+ CC=$LTCC
+ else
+ CC=${CC-"$CXX"}
+ fi
+ ;;
+*)
+ CC=${CC-"$CXX"}
+ ;;
+esac
CFLAGS=${CFLAGS-"$CXXFLAGS"}
# Allow CC to be a program name with arguments.