Hello list, I've installed newest boost into /usr/local - it's a custom installation and not via emerge/portage.
Today, after upgrading system, I've found that I need to use emerge --update --newuse --deep --with-bdeps=y @world to rebuild some dependencies. In the process of recompiling I first noticed this strange thing: checking for Boost headers version >= 1.36.0... yes checking for Boost's header version... 1_55 Portage installed boost is 1.52, my custom installed one is 1.55. Then after a few seconds I saw this: x86_64-pc-linux-gnu-g++ -DPACKAGE_NAME=\"libixion\" -DPACKAGE_TARNAME=\"libixion\" - DPACKAGE_VERSION=\"0.5.0\" -DPACKAGE_STRING=\"libixion\ 0.5.0\" -DPACKAGE_BUGREPORT=\"\" -DPACKAGE_URL=\"\" -DPACKAGE=\"libixion\" -DVERSION=\"0.5.0\" -DSTDC_HEADERS=1 - DHAVE_SYS_TYPES_H=1 -DHAVE_SYS_STAT_H=1 -DHAVE_STDLIB_H=1 -DHAVE_STRING_H=1 - DHAVE_MEMORY_H=1 -DHAVE_STRINGS_H=1 -DHAVE_INTTYPES_H=1 -DHAVE_STDINT_H=1 - DHAVE_UNISTD_H=1 -DHAVE_DLFCN_H=1 -DLT_OBJDIR=\".libs/\" -DHAVE_STDLIB_H=1 - DHAVE_SYS_TIME_H=1 -DHAVE_UNISTD_H=1 -DHAVE__BOOL=1 -DHAVE_STDBOOL_H=1 - DHAVE_GETTIMEOFDAY=1 -DHAVE_BOOST=1 -DHAVE_BOOST_UNORDERED_MAP_HPP=1 - DHAVE_MDDS_RECTANGLE_SET_HPP=1 -DHAVE_MDDS_MIXED_TYPE_MATRIX_HPP=1 - DHAVE_MDDS_MULTI_TYPE_VECTOR_TRAIT_HPP=1 -DHAVE_BOOST_SYSTEM_ERROR_CODE_HPP=1 - DHAVE_BOOST_THREAD_HPP=1 -DHAVE_BOOST_PROGRAM_OPTIONS_HPP=1 -I. -I../include - I../lib/libixion/libixion.la -D_REENTRANT -DMDDS_HASH_CONTAINER_BOOST - D__IXION_BUILDING_DLL -g -Os -fvisibility=hidden -O2 -pipe -march=native -c -o ixion_sorter- sort_input_parser.o `test -f 'sort_input_parser.cpp' || echo './'`sort_input_parser.cpp /bin/sh ../libtool --tag=CXX --mode=link x86_64-pc-linux-gnu-g++ -O2 -pipe -march=native -Wl,- O1 -Wl,--as-needed -o ixion-parser ixion_parser-ixion_parser.o ixion_parser-model_parser.o libixion/libixion-0.6.la -lboost_thread-mt -lboost_system-mt -pthread -lboost_program_options-mt libtool: link: x86_64-pc-linux-gnu-g++ -O2 -pipe -march=native -Wl,-O1 -Wl,--as-needed -o .libs/ixion-parser ixion_parser-ixion_parser.o ixion_parser-model_parser.o -pthread libixion/.libs/libixion-0.6.so -lboost_thread-mt -lboost_system-mt -lboost_program_options-mt - pthread libixion/.libs/libixion-0.6.so: undefined reference to `boost::thread::start_thread_noexcept()' libixion/.libs/libixion-0.6.so: undefined reference to `boost::thread::join_noexcept()' collect2: error: ld returned 1 exit status make[2]: *** [ixion-parser] Error 1 make[2]: *** Waiting for unfinished jobs.... make[2]: Leaving directory `/var/tmp/portage/dev-libs/libixion-0.5.0/work/libixion-0.5.0/src' make[1]: *** [all-recursive] Error 1 make[1]: Leaving directory `/var/tmp/portage/dev-libs/libixion-0.5.0/work/libixion-0.5.0/src' make: *** [all-recursive] Error 1 * ERROR: dev-libs/libixion-0.5.0::gentoo failed (compile phase): ...so it failed because it tried to use my custom boost... I think. To be sure that this is because of this I moved /usr/local/lib and /usr/local/include to /root and another compilation of this library went fine. After doing so I runned revdep-rebuild and it had to recompile libkolabxml because it was linked to boost in /usr/local/lib. So here is my question: If libreoffice does need boost to compile, and I compiled libreoffice after I installed boost to /usr/local, then why it was able to use correct version of boost (from /usr not /usr/local)? Libreoffice is just an example - there are many other programs that depend on boost, and the boost.thread library is very popular one. libkolabxml at version 1.0.1 libixion at version 0.5.0 Should I create a bug for these two libraries or this is expected behaviour? -- Kacper Kopczyński
