[ros-users] cross-compiling self_test fails

[Daniel Maier]

Hi Brian,

both commands just return a blank line.

Daniel

On Thu, 14 Apr 2011 16:49:53 +0200, Brian Gerkey <gerkey at willowgarage.com>  
wrote:

> What does `rosboost-cfg --lib_dirs` return?
>
> And how about `rospack cflags-only-other self_test`?
>
> 	brian.
>
> On Wed, Apr 13, 2011 at 5:52 AM, Daniel Maier
> <maierd at informatik.uni-freiburg.de> wrote:
>> Hi!
>>
>> I am trying to cross-compile ROS (diamondback) for the Nao (again) in a
>> 32-bit ubuntu maverick chroot environment.
>> However, something seems to go wrong.
>> I have rostoolchain.cmake setup to include to the cross-compilation
>> toolchain file provided by Aldebaran,
>> and manually set CMAKE_FIND_ROOT_PATH to point to the cross-compilation
>> toolchain directories.
>>
>> The ros base system builds without errors, however when I try to compile
>> the package:
>>
>> [ self_test ]
>>
>> I receive the linker error:
>> /usr/lib/libboost_regex-mt.so: undefined reference to
>> `std::ctype<char>::_M_widen_init() const at GLIBCXX_3.4.11'
>> collect2: ld returned 1 exit status
>>
>>
>> I think this problem is that ros tries to link to my local boost version
>> instead of the one provided by Aldebaran.
>> More sepecific, I think the problem is that the file
>> CMakeFiles/selftest_rostest.dir/link.txt
>> contains the passage
>> -Wl,-rpath,/usr/lib
>>
>> If I delete that part from the file, I can compile self_test, but it  
>> gets
>> overwritten once I call rosmake again.
>> The passage is also in
>> test/CMakeFiles/error_selftest.dir/link.txt
>> test/CMakeFiles/no_id_selftest.dir/link.txt
>> test/CMakeFiles/nominal_selftest.dir/link.txt
>> test/CMakeFiles/exception_selftest.dir/link.txt
>>
>> and I think it is somehow automatically generated by the
>> rosbuild_add_gtest_build_flags line (or something like that).
>> However, I was unable to find the macro/script which generates this
>> dependency.
>>
>> Can someone help me out on this and either provide a patch or tell me  
>> what
>> I am potentially doing wrong?
>>
>> Thanks!
>> Dan
>> _______________________________________________
>> ros-users mailing list
>> ros-users at code.ros.org
>> https://code.ros.org/mailman/listinfo/ros-users
>>
> _______________________________________________
> ros-users mailing list
> ros-users at code.ros.org
> https://code.ros.org/mailman/listinfo/ros-users


-- 
Albert-Ludwigs-Universität Freiburg, Humanoid Robots Lab
Contact:   http://www.informatik.uni-freiburg.de/~maierd