Re: Another C++ ABI change?
On 04/15/2015 12:37 PM, Richard Shaw wrote: I take it this is a result of the aforementioned ABI breakage? Linking CXX executable imagespec_test cd /builddir/build/BUILD/oiio-Release-1.5.14/build/linux/src/libOpenImageIO && /usr/bin/cmake -E cmake_link_script CMakeFiles/imagespec_test.dir/link.txt --verbose=1 /usr/bin/c++ -O2 -g -pipe -Wall -Werror=format-security -Wp,-D_FORTIFY_SOURCE=2 -fexceptions -fstack-protector-strong --param=ssp-buffer-size=4 -grecord-gcc-switches -specs=/usr/lib/rpm/redhat/redhat-hardened-cc1 -m64 -mtune=generic -O2 -g -DNDEBUG -Wl,-z,relro -specs=/usr/lib/rpm/redhat/redhat-hardened-ld CMakeFiles/imagespec_test.dir/imagespec_test.cpp.o -o imagespec_test -rdynamic libOpenImageIO.so.1.5.14 -lboost_filesystem -lboost_regex -lboost_system -lboost_thread -lrt -ldl -lOpenColorIO -lopencv_core -lopencv_highgui -lfreetype -lpng -lz -ltiff -ljpeg -lz -ltiff -ljpeg -lField3D -lhdf5 -lopenjpeg -lwebp -lIlmImf -lgif -lHalf -lIex -lImath -lIlmThread -lpthread -lpugixml -Wl,-rpath,/builddir/build/BUILD/oiio-Release-1.5.14/build/linux/src/libOpenImageIO libOpenImageIO.so.1.5.14: undefined reference to `Imf_2_2::Header::name[abi:cxx11]() const' libOpenImageIO.so.1.5.14: undefined reference to `Imf_2_2::Header::type[abi:cxx11]() const' collect2: error: ld returned 1 exit status Thanks, Richard Yes, whatever provides "Imf_2_2" needs to get rebuilt. -- Orion Poplawski Technical Manager 303-415-9701 x222 NWRA/CoRA DivisionFAX: 303-415-9702 3380 Mitchell Lane or...@cora.nwra.com Boulder, CO 80301 http://www.cora.nwra.com -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct
Re: Another C++ ABI change?
I take it this is a result of the aforementioned ABI breakage? Linking CXX executable imagespec_test cd /builddir/build/BUILD/oiio-Release-1.5.14/build/linux/src/libOpenImageIO && /usr/bin/cmake -E cmake_link_script CMakeFiles/imagespec_test.dir/link.txt --verbose=1 /usr/bin/c++ -O2 -g -pipe -Wall -Werror=format-security -Wp,-D_FORTIFY_SOURCE=2 -fexceptions -fstack-protector-strong --param=ssp-buffer-size=4 -grecord-gcc-switches -specs=/usr/lib/rpm/redhat/redhat-hardened-cc1 -m64 -mtune=generic -O2 -g -DNDEBUG -Wl,-z,relro -specs=/usr/lib/rpm/redhat/redhat-hardened-ld CMakeFiles/imagespec_test.dir/imagespec_test.cpp.o -o imagespec_test -rdynamic libOpenImageIO.so.1.5.14 -lboost_filesystem -lboost_regex -lboost_system -lboost_thread -lrt -ldl -lOpenColorIO -lopencv_core -lopencv_highgui -lfreetype -lpng -lz -ltiff -ljpeg -lz -ltiff -ljpeg -lField3D -lhdf5 -lopenjpeg -lwebp -lIlmImf -lgif -lHalf -lIex -lImath -lIlmThread -lpthread -lpugixml -Wl,-rpath,/builddir/build/BUILD/oiio-Release-1.5.14/build/linux/src/libOpenImageIO libOpenImageIO.so.1.5.14: undefined reference to `Imf_2_2::Header::name[abi:cxx11]() const' libOpenImageIO.so.1.5.14: undefined reference to `Imf_2_2::Header::type[abi:cxx11]() const' collect2: error: ld returned 1 exit status Thanks, Richard -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct
Re: Another C++ ABI change?
Hi, On Fri, Apr 10, 2015 at 07:26:58AM +0200, Markus Trippelsdorf wrote: > On 2015.04.10 at 00:48 +0200, Michael Schwendt wrote: > > On Thu, 09 Apr 2015 11:40:17 -0600, Orion Poplawski wrote: > > > > > With gcc 5.0.0-0.22.fc23 I'm getting new build failures: > > > > > /builddir/build/BUILD/gdl-0.9.5/src/magick_cl.cpp:175: undefined > > > reference to > > > `Magick::Image::magick[abi:cxx11]() const' > > > > > > Did we just get another set of C++ ABI changes? > > > > I would say "yes", since more C++ libs are affected. > > Jason added automatic ABI tagging for functions with (e.g.) std::string > return type. See: http://thread.gmane.org/gmane.comp.gcc.patches/337867/ > This not only breaks compatibility with clang, but also with all > libraries that were compiled with gcc-5 before this commit. So you have > to rebuild them all again. My interpretation: the gcc team has once again pushed an ABI-breaking change without any announcement and without any intention to handle the necessary rebuilds. Sorry, dear gcc team, but that is not the way we usually do things here... Why didn't you wait for mass rebuild and only built the incompatible gcc into the mass rebuild's tag? Then there would have been no breakage... D. -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct
Re: Another C++ ABI change?
Hi, On Fri, Apr 10, 2015 at 08:33:36AM +0200, Jakub Jelinek wrote: > On Fri, Apr 10, 2015 at 07:26:58AM +0200, Markus Trippelsdorf wrote: > > On 2015.04.10 at 00:48 +0200, Michael Schwendt wrote: > > > On Thu, 09 Apr 2015 11:40:17 -0600, Orion Poplawski wrote: > > > > > > > With gcc 5.0.0-0.22.fc23 I'm getting new build failures: > > > > > > > /builddir/build/BUILD/gdl-0.9.5/src/magick_cl.cpp:175: undefined > > > > reference to > > > > `Magick::Image::magick[abi:cxx11]() const' > > > > > > > > Did we just get another set of C++ ABI changes? > > > > > > I would say "yes", since more C++ libs are affected. > > > > Jason added automatic ABI tagging for functions with (e.g.) std::string > > return type. See: http://thread.gmane.org/gmane.comp.gcc.patches/337867/ > > This not only breaks compatibility with clang, but also with all > > libraries that were compiled with gcc-5 before this commit. So you have > > to rebuild them all again. > > There still will be a F23 mass rebuild, so everything will be rebuilt. That's nice, but meanwhile Rawhide is once again broken. D. -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct
Re: Another C++ ABI change?
On 04/09/2015 11:26 PM, Markus Trippelsdorf wrote: > On 2015.04.10 at 00:48 +0200, Michael Schwendt wrote: >> On Thu, 09 Apr 2015 11:40:17 -0600, Orion Poplawski wrote: >> >>> With gcc 5.0.0-0.22.fc23 I'm getting new build failures: >> >>> /builddir/build/BUILD/gdl-0.9.5/src/magick_cl.cpp:175: undefined reference >>> to >>> `Magick::Image::magick[abi:cxx11]() const' >>> >>> Did we just get another set of C++ ABI changes? >> >> I would say "yes", since more C++ libs are affected. > > Jason added automatic ABI tagging for functions with (e.g.) std::string > return type. See: http://thread.gmane.org/gmane.comp.gcc.patches/337867/ > This not only breaks compatibility with clang, but also with all > libraries that were compiled with gcc-5 before this commit. So you have > to rebuild them all again. > Thanks for the info. I wonder what the chance is of this changing again or being reverted based on this issue. -- Orion Poplawski Technical Manager 303-415-9701 x222 NWRA, Boulder/CoRA Office FAX: 303-415-9702 3380 Mitchell Lane or...@nwra.com Boulder, CO 80301 http://www.nwra.com -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct
Another C++ ABI change?
With gcc 5.0.0-0.22.fc23 I'm getting new build failures: http://koschei.cloud.fedoraproject.org/package/gdl CMakeFiles/gdl.dir/CFMTLexer.cpp.o: In function `antlr::TokenStreamRecognitionException::toString[abi:cxx11]() const': /usr/include/antlr/TokenStreamRecognitionException.hpp:34: undefined reference to `antlr::RecognitionException::getFileLineColumnString[abi:cxx11]() const' CMakeFiles/gdl.dir/FMTIn.cpp.o:(.data.rel.ro._ZTVN5antlr7BaseASTE[_ZTVN5antlr7BaseASTE]+0xd8): undefined reference to `antlr::BaseAST::toStringList[abi:cxx11]() const' CMakeFiles/gdl.dir/FMTIn.cpp.o:(.data.rel.ro._ZTVN5antlr7BaseASTE[_ZTVN5antlr7BaseASTE]+0xe0): undefined reference to `antlr::BaseAST::toStringTree[abi:cxx11]() const' CMakeFiles/gdl.dir/GDLLexer.cpp.o:(.data.rel.ro._ZTVN5antlr17SemanticExceptionE[_ZTVN5antlr17SemanticExceptionE]+0x20): undefined reference to `antlr::RecognitionException::toString[abi:cxx11]() const' CMakeFiles/gdl.dir/GDLLexer.cpp.o:(.data.rel.ro._ZTVN5antlr17SemanticExceptionE[_ZTVN5antlr17SemanticExceptionE]+0x48): undefined reference to `antlr::RecognitionException::getFileLineColumnString[abi:cxx11]() const' CMakeFiles/gdl.dir/dnode.cpp.o:(.data.rel.ro._ZTVN5antlr9CommonASTE[_ZTVN5antlr9CommonASTE]+0xd8): undefined reference to `antlr::BaseAST::toStringList[abi:cxx11]() const' CMakeFiles/gdl.dir/dnode.cpp.o:(.data.rel.ro._ZTVN5antlr9CommonASTE[_ZTVN5antlr9CommonASTE]+0xe0): undefined reference to `antlr::BaseAST::toStringTree[abi:cxx11]() const' CMakeFiles/gdl.dir/dnode.cpp.o:(.data.rel.ro._ZTV5DNode[_ZTV5DNode]+0xd8): undefined reference to `antlr::BaseAST::toStringList[abi:cxx11]() const' CMakeFiles/gdl.dir/dnode.cpp.o:(.data.rel.ro._ZTV5DNode[_ZTV5DNode]+0xe0): undefined reference to `antlr::BaseAST::toStringTree[abi:cxx11]() const' CMakeFiles/gdl.dir/fmtnode.cpp.o:(.data.rel.ro._ZTV7FMTNode[_ZTV7FMTNode]+0xd8): undefined reference to `antlr::BaseAST::toStringList[abi:cxx11]() const' CMakeFiles/gdl.dir/fmtnode.cpp.o:(.data.rel.ro._ZTV7FMTNode[_ZTV7FMTNode]+0xe0): undefined reference to `antlr::BaseAST::toStringTree[abi:cxx11]() const' CMakeFiles/gdl.dir/magick_cl.cpp.o: In function `lib::magick_magick(EnvT*)': /builddir/build/BUILD/gdl-0.9.5/src/magick_cl.cpp:767: undefined reference to `Magick::Image::magick[abi:cxx11]() const' CMakeFiles/gdl.dir/magick_cl.cpp.o: In function `lib::magick_ping(EnvT*)': /builddir/build/BUILD/gdl-0.9.5/src/magick_cl.cpp:231: undefined reference to `Magick::Image::magick[abi:cxx11]() const' /builddir/build/BUILD/gdl-0.9.5/src/magick_cl.cpp:232: undefined reference to `Magick::Image::magick[abi:cxx11]() const' /builddir/build/BUILD/gdl-0.9.5/src/magick_cl.cpp:231: undefined reference to `Magick::Image::magick[abi:cxx11]() const' /builddir/build/BUILD/gdl-0.9.5/src/magick_cl.cpp:175: undefined reference to `Magick::Image::magick[abi:cxx11]() const' Did we just get another set of C++ ABI changes? -- Orion Poplawski Technical Manager 303-415-9701 x222 NWRA, Boulder/CoRA Office FAX: 303-415-9702 3380 Mitchell Lane or...@nwra.com Boulder, CO 80301 http://www.nwra.com -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct