New submission from Max Cantor: On certain Linux distributions such as Ubuntu, the linker is invoked by default with --as-needed, which has an undesireable side effect when linking static libraries: it is bad at detecting required symbols, and the order of libraries on the command line become significant.
Right now, on my Ubuntu 12.10 system with a custom 32-bit version of Python, I get the following command output: mcantor@hpmongo:~$ /opt/pym32/bin/python-config --ldflags -L/opt/pym32/lib/python2.7/config -lpthread -ldl -lutil -lm -lpython2.7 -Xlinker -export-dynamic When linking a project with those flags, I get the following error: /usr/bin/ld: /opt/pym32/lib/python2.7/config/libpython2.7.a(dynload_shlib.o): undefined reference to symbol 'dlopen@@GLIBC_2.1' /usr/bin/ld: note: 'dlopen@@GLIBC_2.1' is defined in DSO /usr/lib/gcc/x86_64-linux-gnu/4.7/../../../i386-linux-gnu/libdl.so so try adding it to the linker command line /usr/lib/gcc/x86_64-linux-gnu/4.7/../../../i386-linux-gnu/libdl.so: could not read symbols: Invalid operation collect2: error: ld returned 1 exit status To resolve the error, I moved -ldl and -lutil *AFTER* -lpython2.7, so the relevant chunk of my gcc command line looked like this: -L/opt/pym32/lib/python2.7/config -lpthread -lm -lpython2.7 -ldl -lutil -Xlinker -export-dynamic I have no idea why --as-needed has such an unpleasant side effect when static libraries are being used, and it's arguable from my perspective that this behavior is the real bug. However it's equally likely that there's a good reason for that behavior, like it causes a slowdown during leap-years on Apple IIs or something. So here I am. python-config ought to respect the quirks of --as-needed when outputting its ldflags. ---------- components: Build, Cross-Build messages: 187121 nosy: Max.Cantor priority: normal severity: normal status: open title: python-config --ldflags gives broken output when statically linking Python with --as-needed type: behavior versions: Python 2.7 _______________________________________ Python tracker <rep...@bugs.python.org> <http://bugs.python.org/issue17769> _______________________________________ _______________________________________________ Python-bugs-list mailing list Unsubscribe: http://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com
