site_archiver@lists.apple.com Delivered-To: darwin-dev@lists.apple.com Organization: University of California San Francisco--Biostatistics The loader is complaining that some object files have multiply defined symbols, even though the symbol in question is defined inside an unnamed (anonymous) namespace and thus should be local to the translation unit by the C++ standard. I don't get this problem on other systems. I worked around this with the -m flag to the linker ("This flag's use is strongly discouraged" says the man page). Is there a better solution? Is this a known problem? Details: The software has not been refreshed recently, so is a little behind: $ uname -a Darwin statcluster 7.9.0 Darwin Kernel Version 7.9.0: Wed Mar 30 20:11:17 PST 2005; root:xnu/xnu-517.12.7.obj~1/RELEASE_PPC Power Macintosh powerpc $ g++ --version g++ (GCC) 3.3 20030304 (Apple Computer, Inc. build 1671) Extract of command line (with many .o files removed for brevity): # That's GNU libtool, not Darwin's ../libtool --mode=link --tag=CXX g++ -o test1 -g \ AbstractTimeStepsGenerator.o Coefficients.o \ -lboost_unit_test_framework -L/usr/local/lib/boost-1_33_1 # It produces the following command g++ -o test1 -g AbstractTimeStepsGenerator.o Coefficients.o Covariates.o -Wl,-bind_\ at_load -lboost_unit_test_framework -L/usr/local/lib/boost-1_33_1 Representative errors: ld: multiple definitions of symbol boost::test_tools::(anonymous namespace)::check_is_close mspathCEntry_test.o definition of boost::test_tools::(anonymous namespace)::check_is_close in section (__DATA,__data) ModelBuilder_test.o definition of boost::test_tools::(anonymous namespace)::check_is_close in section (__DATA,__data) ld: multiple definitions of symbol boost::test_tools::(anonymous namespace)::check_is_small mspathCEntry_test.o definition of boost::test_tools::(anonymous namespace)::check_is_small in section (__DATA,__data) ModelBuilder_test.o definition of boost::test_tools::(anonymous namespace)::check_is_small in section (__DATA,__data) LinearProduct_test.o definition of boost::test_tools::(anonymous namespace)::check_is_close in section (__DATA,__data) LinearProduct_test.o definition of boost::test_tools::(anonymous namespace)::check_is_small in section (__DATA,__data) I have attempted to isolate the problem, but it has failed to reappear in small test cases. Here's my shortest recipe, which relies on Boost 1.33: --------------------- Makefile ------------------------------- CPPFLAGS += -I/usr/local/include/boost-1_33_1 CXXFLAGS += -Wno-long-double LDFLAGS += -L/usr/local/lib/boost-1_33_1 -lboost_unit_test_framework \ -Wl,-m go: t ./t t: a.o b.o ../mspath/src/libtool --mode=link --tag=CXX $(CXX) -o $@ $^ \ $(LDFLAGS) clean: rm *.o t ------------------------ a.cc------------------------------- #define BOOST_AUTO_TEST_MAIN #include <boost/test/auto_unit_test.hpp> #include <boost/test/test_tools.hpp> #include <boost/test/floating_point_comparison.hpp> BOOST_AUTO_UNIT_TEST(a){ BOOST_CHECK_CLOSE(4.1, 4.15, 20); } ------------------------- b.cc ------------------------ #include <boost/test/auto_unit_test.hpp> #include <boost/test/test_tools.hpp> #include <boost/test/floating_point_comparison.hpp> BOOST_AUTO_UNIT_TEST(b){ BOOST_CHECK_CLOSE(3.0, 3.0, .1); } ---------------------------------------- Here's a piece of floating_point_comparison.hpp that is related to the error: struct check_is_close_t { // Public typedefs typedef bool result_type; template<typename FPT, typename PersentType> bool operator()( FPT left, FPT right, PersentType percentage_tolerance, floating_point_comparison_type fpc_type = FPC_STRONG ) { close_at_tolerance<FPT,PersentType> pred( percentage_tolerance, fpc_type ); return pred( left, right ); } }; // the actual error refers to the check_is_close object below namespace { check_is_close_t check_is_close; } -- Ross Boylan wk: (415) 514-8146 185 Berry St #5700 ross@biostat.ucsf.edu Dept of Epidemiology and Biostatistics fax: (415) 514-8150 University of California, San Francisco San Francisco, CA 94107-1739 hm: (415) 550-1062 _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev@lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.appl... This email sent to site_archiver@lists.apple.com