Hi, Domenico Andreoli wrote:
> the main part of boost libraries builds correctly without the -mlong-calls > switch, which worked around this bug. instead a similar problem is now > present in the build of Boost.Graph library, which again is worked around > adding -mlong-calls. > > i recently added the compilation and linking of a missing file [0] > to Boost.Graph, which is right where the new problem arises. hence i'm > not really able to say whether it is really new or simply it was ignored. > > this is the compilation error: > > g++ -ftemplate-depth-50 -shared -Wl,-soname [etc etc] > /usr/bin/ld: read_graphviz_spirit.o(.gnu.linkonce.t._ZNK7 [...] > const&) const]+0xac): cannot reach 00000fec__ZN7phoenix [...] > tuple_indexIXT_EEE+0, recompile with -ffunction-sections > /usr/bin/ld: read_graphviz_spirit.o(.gnu.linkonce.t._ZNK7phoenix14 [...] > phoenix::nil_t> const&) const]+0xac): cannot handle > R_PARISC_PCREL17F for phoenix::tuple_element<0, [...] > >::operator[]<0>(phoenix::tuple_index<0>) > /usr/bin/ld: final link failed: Bad value Matthias Klose wrote: > please could you recheck with current binutils, gcc-4.3, gcc-4.4 in unstable? Ping. It would be nice to be able to at least fix the cryptic error message in binutils upstream. If someone with access to an hppa were to rebuild boost using g++-4.6 and recent binutils and - if it builds successfully, report back - if the linker fails, put up a tarball with the relevant compiled object files and a linker commandline somewhere and report back that would be very helpful. -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
