Bug#710337: libitpp: FTBFS on armhf - multiple typeinfo definitions linking itpp_gtests

Kumar Appaiah a.kumar at alumni.iitm.ac.in
Sat Jun 1 19:48:39 UTC 2013


Hi Aaron.

On Wed, May 29, 2013 at 09:48:32PM -0400, Aaron M. Ucko wrote:
> The armhf build of libitpp failed when trying to link itpp_gtests,
> with reports of multiply defined typeinfo nodes:
> 
>   [...]
>   CMakeFiles/itpp_gtests.dir/multilateration_test.cpp.o:(.rodata+0x70): multiple definition of `typeinfo name for itpp::Array<itpp::Fading_Generator*>'
>   CMakeFiles/itpp_gtests.dir/commfunc_test.cpp.o:(.rodata+0x60): first defined here
>   CMakeFiles/itpp_gtests.dir/multilateration_test.cpp.o:(.rodata+0x98): multiple definition of `typeinfo for itpp::Array<itpp::Fading_Generator*>'
>   CMakeFiles/itpp_gtests.dir/commfunc_test.cpp.o:(.rodata+0x88): first defined here
>   collect2: ld returned 1 exit status
>   make[3]: *** [gtests/itpp_gtests] Error 1
>   make[2]: *** [gtests/CMakeFiles/itpp_gtests.dir/all] Error 2
>   make[3]: Leaving directory `/«PKGBUILDDIR»/build'
>   make[1]: *** [all] Error 2
> 
> I'm not sure why these are clashing, as I would have expected them to
> be safely weak; perhaps it's a compiler bug, in which case you could
> try conditionally build-depending on, and then explicitly using,
> g++-4.7 or g++-4.8.

Thanks for the bug report. This is surprising to me as well, since
this error doesn't manifest on other architectures. I have changed the
build system, I will investigate this on a porterbox, if possible, to
determine the cause.

Thanks.

Kumar
-- 
Kumar Appaiah



More information about the debian-science-maintainers mailing list