Re: [ros-users] Error building cminpack in pcl

classic Classic list List threaded Threaded
8 messages Options
Reply | Threaded
Open this post in threaded view
|

Re: [ros-users] Error building cminpack in pcl

Radu B. Rusu
Administrator
Nicholas,

What version, OS, inside/outside ROS, etc? :)

Cheers,
Radu.


On 12/02/2010 11:45 AM, Nicholas Butko wrote:

> In trying to compile pcl, I got an error from cminpack. The error reads
>
> echo "Building CMinpack..."
> Building CMinpack...
> cd build/cminpack-1.1.0&&  mkdir -p build&&  cd build&&  cmake -DUSE_FPIC=ON -DSHARED_LIBS=ON -DBUILD_EXAMPLES=OFF ..&&  make&&  cd ../../../../
> -- Operating system is generic Unix
> -- Building shared libraries.
> CMake Error at CMakeLists.txt:57 (install):
>    install TARGETS given no LIBRARY DESTINATION for shared library target
>    "cminpack".
>
>
> -- Configuring incomplete, errors occurred!
> make: *** [installed] Error 1
> Granite:cminpack nick$
>
> The relevant line reads:
>
> install (TARGETS cminpack LIBRARY DESTINATION ${CMINPACK_LIB_INSTALL_DIR}
>      ARCHIVE DESTINATION ${CMINPACK_LIB_INSTALL_DIR} COMPONENT library)
>
>
> Maybe CMINPACK_LIB_INSTALL_DIR needs to be set in the make script?
> _______________________________________________
> ros-users mailing list
> [hidden email]
> https://code.ros.org/mailman/listinfo/ros-users
_______________________________________________
[hidden email] / http://pcl.ros.org
https://code.ros.org/mailman/listinfo/pcl-users
Reply | Threaded
Open this post in threaded view
|

Re: [ros-users] Error building cminpack in pcl

Rich Mattes
cmake/cminpack_utils.cmake only defines the library install dir on linux
systems.  The below patch defines "lib" as the library install directory for
all non-linux operating systems.

Rich

--- ./cmake/cminpack_utils.cmake.orig 2010-12-02 15:05:36.568756300 -0500
+++ ./cmake/cminpack_utils.cmake 2010-12-02 15:15:02.360668300 -0500
@@ -4,6 +4,8 @@
     if(WIN32)
         set(OS_WIN TRUE)
     endif(WIN32)
+    
+    set(CMINPACK_LIB_INSTALL_DIR "lib")
     if(OS_LINUX)
         if(${CMAKE_SYSTEM_PROCESSOR} STREQUAL "x86_64")
             set(CMINPACK_LIB_INSTALL_DIR "lib64")


> -----Original Message-----
> From: [hidden email] [mailto:pcl-users-
> [hidden email]] On Behalf Of Radu Bogdan Rusu
> Sent: Thursday, December 02, 2010 2:52 PM
> To: Point Cloud Library (PCL) mailing list
> Cc: Nicholas Butko
> Subject: Re: [PCL-users] [ros-users] Error building cminpack in pcl
>
> Nicholas,
>
> What version, OS, inside/outside ROS, etc? :)
>
> Cheers,
> Radu.
>
>
> On 12/02/2010 11:45 AM, Nicholas Butko wrote:
> > In trying to compile pcl, I got an error from cminpack. The error
> reads
> >
> > echo "Building CMinpack..."
> > Building CMinpack...
> > cd build/cminpack-1.1.0&&  mkdir -p build&&  cd build&&  cmake -
> DUSE_FPIC=ON -DSHARED_LIBS=ON -DBUILD_EXAMPLES=OFF ..&&  make&&  cd
> ../../../../
> > -- Operating system is generic Unix
> > -- Building shared libraries.
> > CMake Error at CMakeLists.txt:57 (install):
> >    install TARGETS given no LIBRARY DESTINATION for shared library
> target
> >    "cminpack".
> >
> >
> > -- Configuring incomplete, errors occurred!
> > make: *** [installed] Error 1
> > Granite:cminpack nick$
> >
> > The relevant line reads:
> >
> > install (TARGETS cminpack LIBRARY DESTINATION
> ${CMINPACK_LIB_INSTALL_DIR}
> >      ARCHIVE DESTINATION ${CMINPACK_LIB_INSTALL_DIR} COMPONENT
> library)
> >
> >
> > Maybe CMINPACK_LIB_INSTALL_DIR needs to be set in the make script?
> > _______________________________________________
> > ros-users mailing list
> > [hidden email]
> > https://code.ros.org/mailman/listinfo/ros-users
> _______________________________________________
> [hidden email] / http://pcl.ros.org
> https://code.ros.org/mailman/listinfo/pcl-users

_______________________________________________
[hidden email] / http://pcl.ros.org
https://code.ros.org/mailman/listinfo/pcl-users
Reply | Threaded
Open this post in threaded view
|

Re: [ros-users] Error building cminpack in pcl

Radu B. Rusu
Administrator
Rich,

Thanks. We committed a workaround in the trunk of the cminpack ROS metapackage (as suggested by Nicholas), and we're
waiting for the patch integration and a new release from Fred upstream.

Cheers,
Radu.


On 12/02/2010 12:19 PM, Rich Mattes wrote:

> cmake/cminpack_utils.cmake only defines the library install dir on linux
> systems.  The below patch defines "lib" as the library install directory for
> all non-linux operating systems.
>
> Rich
>
> --- ./cmake/cminpack_utils.cmake.orig 2010-12-02 15:05:36.568756300 -0500
> +++ ./cmake/cminpack_utils.cmake 2010-12-02 15:15:02.360668300 -0500
> @@ -4,6 +4,8 @@
>       if(WIN32)
>           set(OS_WIN TRUE)
>       endif(WIN32)
> +
> +    set(CMINPACK_LIB_INSTALL_DIR "lib")
>       if(OS_LINUX)
>           if(${CMAKE_SYSTEM_PROCESSOR} STREQUAL "x86_64")
>               set(CMINPACK_LIB_INSTALL_DIR "lib64")
>
>
>> -----Original Message-----
>> From: [hidden email] [mailto:pcl-users-
>> [hidden email]] On Behalf Of Radu Bogdan Rusu
>> Sent: Thursday, December 02, 2010 2:52 PM
>> To: Point Cloud Library (PCL) mailing list
>> Cc: Nicholas Butko
>> Subject: Re: [PCL-users] [ros-users] Error building cminpack in pcl
>>
>> Nicholas,
>>
>> What version, OS, inside/outside ROS, etc? :)
>>
>> Cheers,
>> Radu.
>>
>>
>> On 12/02/2010 11:45 AM, Nicholas Butko wrote:
>>> In trying to compile pcl, I got an error from cminpack. The error
>> reads
>>>
>>> echo "Building CMinpack..."
>>> Building CMinpack...
>>> cd build/cminpack-1.1.0&&   mkdir -p build&&   cd build&&   cmake -
>> DUSE_FPIC=ON -DSHARED_LIBS=ON -DBUILD_EXAMPLES=OFF ..&&   make&&   cd
>> ../../../../
>>> -- Operating system is generic Unix
>>> -- Building shared libraries.
>>> CMake Error at CMakeLists.txt:57 (install):
>>>     install TARGETS given no LIBRARY DESTINATION for shared library
>> target
>>>     "cminpack".
>>>
>>>
>>> -- Configuring incomplete, errors occurred!
>>> make: *** [installed] Error 1
>>> Granite:cminpack nick$
>>>
>>> The relevant line reads:
>>>
>>> install (TARGETS cminpack LIBRARY DESTINATION
>> ${CMINPACK_LIB_INSTALL_DIR}
>>>       ARCHIVE DESTINATION ${CMINPACK_LIB_INSTALL_DIR} COMPONENT
>> library)
>>>
>>>
>>> Maybe CMINPACK_LIB_INSTALL_DIR needs to be set in the make script?
>>> _______________________________________________
>>> ros-users mailing list
>>> [hidden email]
>>> https://code.ros.org/mailman/listinfo/ros-users
>> _______________________________________________
>> [hidden email] / http://pcl.ros.org
>> https://code.ros.org/mailman/listinfo/pcl-users
>
> _______________________________________________
> [hidden email] / http://pcl.ros.org
> https://code.ros.org/mailman/listinfo/pcl-users
_______________________________________________
[hidden email] / http://pointclouds.org
https://code.ros.org/mailman/listinfo/pcl-users
Reply | Threaded
Open this post in threaded view
|

Re: [ros-users] Error building cminpack in pcl

Radu B. Rusu
Administrator
In reply to this post by Radu B. Rusu
CC-ing back the list, maybe some boost macos experts/users can help us identify this issue.

So, the OpenMP thing should not be a big problem. If they're just warnings you can ignore them. In principle, the
compiler should support this -- but I have no experience with MacOS.

What about the Boost issues? Patrick do you happen to know where they might come from?

Cheers,
Radu.


On 12/03/2010 12:37 PM, Nicholas Butko wrote:

> Jeez. Boost takes a long time to build.
>
> Anyway, the problem did not fix itself. My boost version is 1.45.0_1, my OS is OSX 10.6.5. The perception_pcl stack I am using is "unstable".
>
> If you have any questions or suggestions, I can try things out for you.
>
> By the way, in an unrelated issue, I also get lots of warnings related to openmp, such as
>
> -- Looking for omp_set_num_threads
> -- Looking for omp_set_num_threads - not found
> -- Looking for omp_set_num_threads
> -- Looking for omp_set_num_threads - not found
> -- Looking for omp_set_num_threads
> -- Looking for omp_set_num_threads - not found
> -- Looking for omp_set_num_threads
> -- Looking for omp_set_num_threads - not found
> -- Looking for omp_set_num_threads
> -- Looking for omp_set_num_threads - not found
> -- Looking for omp_set_num_threads
> -- Looking for omp_set_num_threads - not found
> WARNING: OpenMP compile flag not found
> WARNING: OpenMP compile flag not found
> WARNING: OpenMP compile flag not found
> WARNING: OpenMP compile flag not found
> WARNING: OpenMP compile flag not found
>
>
> On Dec 3, 2010, at 10:48 AM, Radu Bogdan Rusu wrote:
>
>> Whoops. Ok, now I really have to know what OS, inside/outside ROS.... :)
>>
>> Cheers,
>> Radu.
>>
>>
>> On 12/03/2010 10:45 AM, Nicholas Butko wrote:
>>> No problem.
>>>
>>> Unfortunately, now I'm having trouble building pcl. The problem seems to be with POINT_CLOUD_REGISTER_POINT_STRUCT(). I get errors like
>>>
>>> In file included from /Users/nick/ros/stacks/perception_pcl/pcl/include/pcl/range_image/range_image.h:45,
>>>                   from /Users/nick/ros/stacks/perception_pcl/pcl/src/pcl/range_image/range_image.cpp:46:
>>> /Users/nick/ros/stacks/perception_pcl/pcl/include/pcl/point_types.h:188: error: expected identifier before ‘(’ token
>>> /Users/nick/ros/stacks/perception_pcl/pcl/include/pcl/point_types.h:188: error: expected unqualified-id before ‘,’ token
>>> /Users/nick/ros/stacks/perception_pcl/pcl/include/pcl/point_types.h:188: error: expected `)' before ‘,’ token
>>> /Users/nick/ros/stacks/perception_pcl/pcl/include/pcl/point_types.h:188: error: expected `)' before ‘,’ token
>>> /Users/nick/ros/stacks/perception_pcl/pcl/include/pcl/point_types.h:188: error: ‘x’ has not been declared
>>> /Users/nick/ros/stacks/perception_pcl/pcl/include/pcl/point_types.h:188: error: expected constructor, destructor, or type conversion before ‘)’ token
>>>
>>> on every line with POINT_CLOUD_REGISTER_POINT_STRUCT. The POINT_CLOUD_REGISTER_POINT_WRAPPER does not seem to be a problem.
>>>
>>> --Nick
>>>
>>> On Dec 3, 2010, at 10:18 AM, Radu Bogdan Rusu wrote:
>>>
>>>> Thanks Nicholas! I pushed the patch in trunk.
>>>>
>>>> http://www.ros.org/wiki/perception_pcl/ChangeList
>>>>
>>>> Cheers,
>>>> Radu.
>>>>
>>>>
>>>> On 12/02/2010 03:35 PM, Nicholas Butko wrote:
>>>>> Radu,
>>>>>
>>>>> The attached patch works for me. The only change is to add -DCMINPACK_LIB_INSTALL_DIR="." to the cmake call.
>>>>>
>>>>> --Nick
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>> On Dec 2, 2010, at 11:51 AM, Radu Bogdan Rusu wrote:
>>>>>
>>>>>> Nicholas,
>>>>>>
>>>>>> What version, OS, inside/outside ROS, etc? :)
>>>>>>
>>>>>> Cheers,
>>>>>> Radu.
>>>>>>
>>>>>>
>>>>>> On 12/02/2010 11:45 AM, Nicholas Butko wrote:
>>>>>>> In trying to compile pcl, I got an error from cminpack. The error reads
>>>>>>>
>>>>>>> echo "Building CMinpack..."
>>>>>>> Building CMinpack...
>>>>>>> cd build/cminpack-1.1.0&&     mkdir -p build&&     cd build&&     cmake -DUSE_FPIC=ON -DSHARED_LIBS=ON -DBUILD_EXAMPLES=OFF ..&&     make&&     cd ../../../../
>>>>>>> -- Operating system is generic Unix
>>>>>>> -- Building shared libraries.
>>>>>>> CMake Error at CMakeLists.txt:57 (install):
>>>>>>>    install TARGETS given no LIBRARY DESTINATION for shared library target
>>>>>>>    "cminpack".
>>>>>>>
>>>>>>>
>>>>>>> -- Configuring incomplete, errors occurred!
>>>>>>> make: *** [installed] Error 1
>>>>>>> Granite:cminpack nick$
>>>>>>>
>>>>>>> The relevant line reads:
>>>>>>>
>>>>>>> install (TARGETS cminpack LIBRARY DESTINATION ${CMINPACK_LIB_INSTALL_DIR}
>>>>>>>      ARCHIVE DESTINATION ${CMINPACK_LIB_INSTALL_DIR} COMPONENT library)
>>>>>>>
>>>>>>>
>>>>>>> Maybe CMINPACK_LIB_INSTALL_DIR needs to be set in the make script?
>>>>>>> _______________________________________________
>>>>>>> ros-users mailing list
>>>>>>> [hidden email]
>>>>>>> https://code.ros.org/mailman/listinfo/ros-users
>>>>>
>>>
>
_______________________________________________
[hidden email] / http://pointclouds.org
https://code.ros.org/mailman/listinfo/pcl-users
Reply | Threaded
Open this post in threaded view
|

Error building pcl related to boost fusion

Nicholas Butko
Radu,

I finally had some time to sit down with this one, and I think I figured it out. However, someone much more familiar with boost than me should confirm this.

It seems like "register_point_struct.h" uses a preprocessor macro that no longer exists in boost/fusion, specifically BOOST_FUSION_ADAPT_ASSOC_STRUCT_I. When I ran point_types.h through the preprocessor, this didn't get resolved. When I looked through my boost library (1.45), it didn't exist. When I looked through an older boost library (1.41) it did.

So the next step was to confirm if it was relegated to macports or more general. I downloaded the boost source from boost.org and sure enough, the macro wasn't defined in the current version of boost. So it's not just a problem with macports.

I am attaching a patch that I hope fixes the problem temporarily.  Unfortunately, it does this by calling undocumented macros like BOOST_FUSION_ADAPT_ASSOC_STRUCT_BASE and BOOST_FUSION_ADAPT_ASSOC_STRUCT_C. It would probably be best if register_point_struct.h could be rewritten to use only documented macros, like BOOST_FUSTION_ADAPT_ASSOC_STRUCT.

I am not that confident in the patch, because basically I tried to guess what was going on from the implementation of BOOST_FUSTION_ADAPT_ASSOC_STRUCT. Someone should look it over to see if it's doing the right thing, and to make sure it doesn't break anything.

--Nick



 

On Dec 3, 2010, at 1:03 PM, Radu Bogdan Rusu wrote:

> CC-ing back the list, maybe some boost macos experts/users can help us identify this issue.
>
> So, the OpenMP thing should not be a big problem. If they're just warnings you can ignore them. In principle, the compiler should support this -- but I have no experience with MacOS.
>
> What about the Boost issues? Patrick do you happen to know where they might come from?
>
> Cheers,
> Radu.
>
>
> On 12/03/2010 12:37 PM, Nicholas Butko wrote:
>> Jeez. Boost takes a long time to build.
>>
>> Anyway, the problem did not fix itself. My boost version is 1.45.0_1, my OS is OSX 10.6.5. The perception_pcl stack I am using is "unstable".
>>
>> If you have any questions or suggestions, I can try things out for you.
>>
>> By the way, in an unrelated issue, I also get lots of warnings related to openmp, such as
>>
>> -- Looking for omp_set_num_threads
>> -- Looking for omp_set_num_threads - not found
>> -- Looking for omp_set_num_threads
>> -- Looking for omp_set_num_threads - not found
>> -- Looking for omp_set_num_threads
>> -- Looking for omp_set_num_threads - not found
>> -- Looking for omp_set_num_threads
>> -- Looking for omp_set_num_threads - not found
>> -- Looking for omp_set_num_threads
>> -- Looking for omp_set_num_threads - not found
>> -- Looking for omp_set_num_threads
>> -- Looking for omp_set_num_threads - not found
>> WARNING: OpenMP compile flag not found
>> WARNING: OpenMP compile flag not found
>> WARNING: OpenMP compile flag not found
>> WARNING: OpenMP compile flag not found
>> WARNING: OpenMP compile flag not found
>>
>>
>> On Dec 3, 2010, at 10:48 AM, Radu Bogdan Rusu wrote:
>>
>>> Whoops. Ok, now I really have to know what OS, inside/outside ROS.... :)
>>>
>>> Cheers,
>>> Radu.
>>>
>>>
>>> On 12/03/2010 10:45 AM, Nicholas Butko wrote:
>>>> No problem.
>>>>
>>>> Unfortunately, now I'm having trouble building pcl. The problem seems to be with POINT_CLOUD_REGISTER_POINT_STRUCT(). I get errors like
>>>>
>>>> In file included from /Users/nick/ros/stacks/perception_pcl/pcl/include/pcl/range_image/range_image.h:45,
>>>>                  from /Users/nick/ros/stacks/perception_pcl/pcl/src/pcl/range_image/range_image.cpp:46:
>>>> /Users/nick/ros/stacks/perception_pcl/pcl/include/pcl/point_types.h:188: error: expected identifier before ‘(’ token
>>>> /Users/nick/ros/stacks/perception_pcl/pcl/include/pcl/point_types.h:188: error: expected unqualified-id before ‘,’ token
>>>> /Users/nick/ros/stacks/perception_pcl/pcl/include/pcl/point_types.h:188: error: expected `)' before ‘,’ token
>>>> /Users/nick/ros/stacks/perception_pcl/pcl/include/pcl/point_types.h:188: error: expected `)' before ‘,’ token
>>>> /Users/nick/ros/stacks/perception_pcl/pcl/include/pcl/point_types.h:188: error: ‘x’ has not been declared
>>>> /Users/nick/ros/stacks/perception_pcl/pcl/include/pcl/point_types.h:188: error: expected constructor, destructor, or type conversion before ‘)’ token
>>>>
>>>> on every line with POINT_CLOUD_REGISTER_POINT_STRUCT. The POINT_CLOUD_REGISTER_POINT_WRAPPER does not seem to be a problem.
>>>>
>>>> --Nick
>>>>
>>>> On Dec 3, 2010, at 10:18 AM, Radu Bogdan Rusu wrote:
>>>>
>>>>> Thanks Nicholas! I pushed the patch in trunk.
>>>>>
>>>>> http://www.ros.org/wiki/perception_pcl/ChangeList
>>>>>
>>>>> Cheers,
>>>>> Radu.
>>>>>
>>>>>
>>>>> On 12/02/2010 03:35 PM, Nicholas Butko wrote:
>>>>>> Radu,
>>>>>>
>>>>>> The attached patch works for me. The only change is to add -DCMINPACK_LIB_INSTALL_DIR="." to the cmake call.
>>>>>>
>>>>>> --Nick
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>> On Dec 2, 2010, at 11:51 AM, Radu Bogdan Rusu wrote:
>>>>>>
>>>>>>> Nicholas,
>>>>>>>
>>>>>>> What version, OS, inside/outside ROS, etc? :)
>>>>>>>
>>>>>>> Cheers,
>>>>>>> Radu.
>>>>>>>
>>>>>>>
>>>>>>> On 12/02/2010 11:45 AM, Nicholas Butko wrote:
>>>>>>>> In trying to compile pcl, I got an error from cminpack. The error reads
>>>>>>>>
>>>>>>>> echo "Building CMinpack..."
>>>>>>>> Building CMinpack...
>>>>>>>> cd build/cminpack-1.1.0&&     mkdir -p build&&     cd build&&     cmake -DUSE_FPIC=ON -DSHARED_LIBS=ON -DBUILD_EXAMPLES=OFF ..&&     make&&     cd ../../../../
>>>>>>>> -- Operating system is generic Unix
>>>>>>>> -- Building shared libraries.
>>>>>>>> CMake Error at CMakeLists.txt:57 (install):
>>>>>>>>   install TARGETS given no LIBRARY DESTINATION for shared library target
>>>>>>>>   "cminpack".
>>>>>>>>
>>>>>>>>
>>>>>>>> -- Configuring incomplete, errors occurred!
>>>>>>>> make: *** [installed] Error 1
>>>>>>>> Granite:cminpack nick$
>>>>>>>>
>>>>>>>> The relevant line reads:
>>>>>>>>
>>>>>>>> install (TARGETS cminpack LIBRARY DESTINATION ${CMINPACK_LIB_INSTALL_DIR}
>>>>>>>>     ARCHIVE DESTINATION ${CMINPACK_LIB_INSTALL_DIR} COMPONENT library)
>>>>>>>>
>>>>>>>>
>>>>>>>> Maybe CMINPACK_LIB_INSTALL_DIR needs to be set in the make script?
>>>>>>>> _______________________________________________
>>>>>>>> ros-users mailing list
>>>>>>>> [hidden email]
>>>>>>>> https://code.ros.org/mailman/listinfo/ros-users
>>>>>>
>>>>
>>

_______________________________________________
[hidden email] / http://pointclouds.org
https://code.ros.org/mailman/listinfo/pcl-users

register_point_struct_h.patch (978 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: Error building pcl related to boost fusion

Nicholas Butko
This patch is simpler, doesn't use undocumented APIs, and does the same thing as the previous patch. However, I'm still not sure if it's doing the *right* thing, particularly because I am getting linking errors now. This is better than before, when I was getting compile errors.

I'm not sure if the linking errors are caused by this patch not doing the right thing, or are caused by something else entirely. This preprocessor stuff is pretty hard to debug, and I'd appreciate if someone with more experience than me could take a look.

--Nick





On Dec 6, 2010, at 10:14 PM, Nicholas Butko wrote:

> Radu,
>
> I finally had some time to sit down with this one, and I think I figured it out. However, someone much more familiar with boost than me should confirm this.
>
> It seems like "register_point_struct.h" uses a preprocessor macro that no longer exists in boost/fusion, specifically BOOST_FUSION_ADAPT_ASSOC_STRUCT_I. When I ran point_types.h through the preprocessor, this didn't get resolved. When I looked through my boost library (1.45), it didn't exist. When I looked through an older boost library (1.41) it did.
>
> So the next step was to confirm if it was relegated to macports or more general. I downloaded the boost source from boost.org and sure enough, the macro wasn't defined in the current version of boost. So it's not just a problem with macports.
>
> I am attaching a patch that I hope fixes the problem temporarily.  Unfortunately, it does this by calling undocumented macros like BOOST_FUSION_ADAPT_ASSOC_STRUCT_BASE and BOOST_FUSION_ADAPT_ASSOC_STRUCT_C. It would probably be best if register_point_struct.h could be rewritten to use only documented macros, like BOOST_FUSTION_ADAPT_ASSOC_STRUCT.
>
> I am not that confident in the patch, because basically I tried to guess what was going on from the implementation of BOOST_FUSTION_ADAPT_ASSOC_STRUCT. Someone should look it over to see if it's doing the right thing, and to make sure it doesn't break anything.
>
> --Nick
>
> <register_point_struct_h.patch>
>
>
> On Dec 3, 2010, at 1:03 PM, Radu Bogdan Rusu wrote:
>
>> CC-ing back the list, maybe some boost macos experts/users can help us identify this issue.
>>
>> So, the OpenMP thing should not be a big problem. If they're just warnings you can ignore them. In principle, the compiler should support this -- but I have no experience with MacOS.
>>
>> What about the Boost issues? Patrick do you happen to know where they might come from?
>>
>> Cheers,
>> Radu.
>>
>>
>> On 12/03/2010 12:37 PM, Nicholas Butko wrote:
>>> Jeez. Boost takes a long time to build.
>>>
>>> Anyway, the problem did not fix itself. My boost version is 1.45.0_1, my OS is OSX 10.6.5. The perception_pcl stack I am using is "unstable".
>>>
>>> If you have any questions or suggestions, I can try things out for you.
>>>
>>> By the way, in an unrelated issue, I also get lots of warnings related to openmp, such as
>>>
>>> -- Looking for omp_set_num_threads
>>> -- Looking for omp_set_num_threads - not found
>>> -- Looking for omp_set_num_threads
>>> -- Looking for omp_set_num_threads - not found
>>> -- Looking for omp_set_num_threads
>>> -- Looking for omp_set_num_threads - not found
>>> -- Looking for omp_set_num_threads
>>> -- Looking for omp_set_num_threads - not found
>>> -- Looking for omp_set_num_threads
>>> -- Looking for omp_set_num_threads - not found
>>> -- Looking for omp_set_num_threads
>>> -- Looking for omp_set_num_threads - not found
>>> WARNING: OpenMP compile flag not found
>>> WARNING: OpenMP compile flag not found
>>> WARNING: OpenMP compile flag not found
>>> WARNING: OpenMP compile flag not found
>>> WARNING: OpenMP compile flag not found
>>>
>>>
>>> On Dec 3, 2010, at 10:48 AM, Radu Bogdan Rusu wrote:
>>>
>>>> Whoops. Ok, now I really have to know what OS, inside/outside ROS.... :)
>>>>
>>>> Cheers,
>>>> Radu.
>>>>
>>>>
>>>> On 12/03/2010 10:45 AM, Nicholas Butko wrote:
>>>>> No problem.
>>>>>
>>>>> Unfortunately, now I'm having trouble building pcl. The problem seems to be with POINT_CLOUD_REGISTER_POINT_STRUCT(). I get errors like
>>>>>
>>>>> In file included from /Users/nick/ros/stacks/perception_pcl/pcl/include/pcl/range_image/range_image.h:45,
>>>>>                 from /Users/nick/ros/stacks/perception_pcl/pcl/src/pcl/range_image/range_image.cpp:46:
>>>>> /Users/nick/ros/stacks/perception_pcl/pcl/include/pcl/point_types.h:188: error: expected identifier before ‘(’ token
>>>>> /Users/nick/ros/stacks/perception_pcl/pcl/include/pcl/point_types.h:188: error: expected unqualified-id before ‘,’ token
>>>>> /Users/nick/ros/stacks/perception_pcl/pcl/include/pcl/point_types.h:188: error: expected `)' before ‘,’ token
>>>>> /Users/nick/ros/stacks/perception_pcl/pcl/include/pcl/point_types.h:188: error: expected `)' before ‘,’ token
>>>>> /Users/nick/ros/stacks/perception_pcl/pcl/include/pcl/point_types.h:188: error: ‘x’ has not been declared
>>>>> /Users/nick/ros/stacks/perception_pcl/pcl/include/pcl/point_types.h:188: error: expected constructor, destructor, or type conversion before ‘)’ token
>>>>>
>>>>> on every line with POINT_CLOUD_REGISTER_POINT_STRUCT. The POINT_CLOUD_REGISTER_POINT_WRAPPER does not seem to be a problem.
>>>>>
>>>>> --Nick
>>>>>
>>>>> On Dec 3, 2010, at 10:18 AM, Radu Bogdan Rusu wrote:
>>>>>
>>>>>> Thanks Nicholas! I pushed the patch in trunk.
>>>>>>
>>>>>> http://www.ros.org/wiki/perception_pcl/ChangeList
>>>>>>
>>>>>> Cheers,
>>>>>> Radu.
>>>>>>
>>>>>>
>>>>>> On 12/02/2010 03:35 PM, Nicholas Butko wrote:
>>>>>>> Radu,
>>>>>>>
>>>>>>> The attached patch works for me. The only change is to add -DCMINPACK_LIB_INSTALL_DIR="." to the cmake call.
>>>>>>>
>>>>>>> --Nick
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> On Dec 2, 2010, at 11:51 AM, Radu Bogdan Rusu wrote:
>>>>>>>
>>>>>>>> Nicholas,
>>>>>>>>
>>>>>>>> What version, OS, inside/outside ROS, etc? :)
>>>>>>>>
>>>>>>>> Cheers,
>>>>>>>> Radu.
>>>>>>>>
>>>>>>>>
>>>>>>>> On 12/02/2010 11:45 AM, Nicholas Butko wrote:
>>>>>>>>> In trying to compile pcl, I got an error from cminpack. The error reads
>>>>>>>>>
>>>>>>>>> echo "Building CMinpack..."
>>>>>>>>> Building CMinpack...
>>>>>>>>> cd build/cminpack-1.1.0&&     mkdir -p build&&     cd build&&     cmake -DUSE_FPIC=ON -DSHARED_LIBS=ON -DBUILD_EXAMPLES=OFF ..&&     make&&     cd ../../../../
>>>>>>>>> -- Operating system is generic Unix
>>>>>>>>> -- Building shared libraries.
>>>>>>>>> CMake Error at CMakeLists.txt:57 (install):
>>>>>>>>>  install TARGETS given no LIBRARY DESTINATION for shared library target
>>>>>>>>>  "cminpack".
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> -- Configuring incomplete, errors occurred!
>>>>>>>>> make: *** [installed] Error 1
>>>>>>>>> Granite:cminpack nick$
>>>>>>>>>
>>>>>>>>> The relevant line reads:
>>>>>>>>>
>>>>>>>>> install (TARGETS cminpack LIBRARY DESTINATION ${CMINPACK_LIB_INSTALL_DIR}
>>>>>>>>>    ARCHIVE DESTINATION ${CMINPACK_LIB_INSTALL_DIR} COMPONENT library)
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> Maybe CMINPACK_LIB_INSTALL_DIR needs to be set in the make script?
>>>>>>>>> _______________________________________________
>>>>>>>>> ros-users mailing list
>>>>>>>>> [hidden email]
>>>>>>>>> https://code.ros.org/mailman/listinfo/ros-users
>>>>>>>
>>>>>
>>>
>

_______________________________________________
[hidden email] / http://pointclouds.org
https://code.ros.org/mailman/listinfo/pcl-users

register_point_struct_h.patch (827 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: Error building pcl related to boost fusion

Patrick Mihelich
Thanks for the report Nick. I should've known the Fusion stuff would bite me :P.

Radu, are we actually using the Fusion registration of point types in PCL at all? I put it in originally to dissociate struct fields from standardized channel names. So for example, you could have some arbitrary type

struct MyPointType
{
  float pos[3];
  // ...
};

And register pos[0], pos[1], pos[2] as channels "x", "y", "z". But if we're always assuming that point.x, point.y, point.z are valid, there's not much purpose.

Patrick

On Mon, Dec 6, 2010 at 11:27 PM, Nicholas Butko <[hidden email]> wrote:
This patch is simpler, doesn't use undocumented APIs, and does the same thing as the previous patch. However, I'm still not sure if it's doing the *right* thing, particularly because I am getting linking errors now. This is better than before, when I was getting compile errors.

I'm not sure if the linking errors are caused by this patch not doing the right thing, or are caused by something else entirely. This preprocessor stuff is pretty hard to debug, and I'd appreciate if someone with more experience than me could take a look.

--Nick





On Dec 6, 2010, at 10:14 PM, Nicholas Butko wrote:

> Radu,
>
> I finally had some time to sit down with this one, and I think I figured it out. However, someone much more familiar with boost than me should confirm this.
>
> It seems like "register_point_struct.h" uses a preprocessor macro that no longer exists in boost/fusion, specifically BOOST_FUSION_ADAPT_ASSOC_STRUCT_I. When I ran point_types.h through the preprocessor, this didn't get resolved. When I looked through my boost library (1.45), it didn't exist. When I looked through an older boost library (1.41) it did.
>
> So the next step was to confirm if it was relegated to macports or more general. I downloaded the boost source from boost.org and sure enough, the macro wasn't defined in the current version of boost. So it's not just a problem with macports.
>
> I am attaching a patch that I hope fixes the problem temporarily.  Unfortunately, it does this by calling undocumented macros like BOOST_FUSION_ADAPT_ASSOC_STRUCT_BASE and BOOST_FUSION_ADAPT_ASSOC_STRUCT_C. It would probably be best if register_point_struct.h could be rewritten to use only documented macros, like BOOST_FUSTION_ADAPT_ASSOC_STRUCT.
>
> I am not that confident in the patch, because basically I tried to guess what was going on from the implementation of BOOST_FUSTION_ADAPT_ASSOC_STRUCT. Someone should look it over to see if it's doing the right thing, and to make sure it doesn't break anything.
>
> --Nick
>
> <register_point_struct_h.patch>
>
>
> On Dec 3, 2010, at 1:03 PM, Radu Bogdan Rusu wrote:
>
>> CC-ing back the list, maybe some boost macos experts/users can help us identify this issue.
>>
>> So, the OpenMP thing should not be a big problem. If they're just warnings you can ignore them. In principle, the compiler should support this -- but I have no experience with MacOS.
>>
>> What about the Boost issues? Patrick do you happen to know where they might come from?
>>
>> Cheers,
>> Radu.
>>
>>
>> On 12/03/2010 12:37 PM, Nicholas Butko wrote:
>>> Jeez. Boost takes a long time to build.
>>>
>>> Anyway, the problem did not fix itself. My boost version is 1.45.0_1, my OS is OSX 10.6.5. The perception_pcl stack I am using is "unstable".
>>>
>>> If you have any questions or suggestions, I can try things out for you.
>>>
>>> By the way, in an unrelated issue, I also get lots of warnings related to openmp, such as
>>>
>>> -- Looking for omp_set_num_threads
>>> -- Looking for omp_set_num_threads - not found
>>> -- Looking for omp_set_num_threads
>>> -- Looking for omp_set_num_threads - not found
>>> -- Looking for omp_set_num_threads
>>> -- Looking for omp_set_num_threads - not found
>>> -- Looking for omp_set_num_threads
>>> -- Looking for omp_set_num_threads - not found
>>> -- Looking for omp_set_num_threads
>>> -- Looking for omp_set_num_threads - not found
>>> -- Looking for omp_set_num_threads
>>> -- Looking for omp_set_num_threads - not found
>>> WARNING: OpenMP compile flag not found
>>> WARNING: OpenMP compile flag not found
>>> WARNING: OpenMP compile flag not found
>>> WARNING: OpenMP compile flag not found
>>> WARNING: OpenMP compile flag not found
>>>
>>>
>>> On Dec 3, 2010, at 10:48 AM, Radu Bogdan Rusu wrote:
>>>
>>>> Whoops. Ok, now I really have to know what OS, inside/outside ROS.... :)
>>>>
>>>> Cheers,
>>>> Radu.
>>>>
>>>>
>>>> On 12/03/2010 10:45 AM, Nicholas Butko wrote:
>>>>> No problem.
>>>>>
>>>>> Unfortunately, now I'm having trouble building pcl. The problem seems to be with POINT_CLOUD_REGISTER_POINT_STRUCT(). I get errors like
>>>>>
>>>>> In file included from /Users/nick/ros/stacks/perception_pcl/pcl/include/pcl/range_image/range_image.h:45,
>>>>>                 from /Users/nick/ros/stacks/perception_pcl/pcl/src/pcl/range_image/range_image.cpp:46:
>>>>> /Users/nick/ros/stacks/perception_pcl/pcl/include/pcl/point_types.h:188: error: expected identifier before ‘(’ token
>>>>> /Users/nick/ros/stacks/perception_pcl/pcl/include/pcl/point_types.h:188: error: expected unqualified-id before ‘,’ token
>>>>> /Users/nick/ros/stacks/perception_pcl/pcl/include/pcl/point_types.h:188: error: expected `)' before ‘,’ token
>>>>> /Users/nick/ros/stacks/perception_pcl/pcl/include/pcl/point_types.h:188: error: expected `)' before ‘,’ token
>>>>> /Users/nick/ros/stacks/perception_pcl/pcl/include/pcl/point_types.h:188: error: ‘x’ has not been declared
>>>>> /Users/nick/ros/stacks/perception_pcl/pcl/include/pcl/point_types.h:188: error: expected constructor, destructor, or type conversion before ‘)’ token
>>>>>
>>>>> on every line with POINT_CLOUD_REGISTER_POINT_STRUCT. The POINT_CLOUD_REGISTER_POINT_WRAPPER does not seem to be a problem.
>>>>>
>>>>> --Nick
>>>>>
>>>>> On Dec 3, 2010, at 10:18 AM, Radu Bogdan Rusu wrote:
>>>>>
>>>>>> Thanks Nicholas! I pushed the patch in trunk.
>>>>>>
>>>>>> http://www.ros.org/wiki/perception_pcl/ChangeList
>>>>>>
>>>>>> Cheers,
>>>>>> Radu.
>>>>>>
>>>>>>
>>>>>> On 12/02/2010 03:35 PM, Nicholas Butko wrote:
>>>>>>> Radu,
>>>>>>>
>>>>>>> The attached patch works for me. The only change is to add -DCMINPACK_LIB_INSTALL_DIR="." to the cmake call.
>>>>>>>
>>>>>>> --Nick
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> On Dec 2, 2010, at 11:51 AM, Radu Bogdan Rusu wrote:
>>>>>>>
>>>>>>>> Nicholas,
>>>>>>>>
>>>>>>>> What version, OS, inside/outside ROS, etc? :)
>>>>>>>>
>>>>>>>> Cheers,
>>>>>>>> Radu.
>>>>>>>>
>>>>>>>>
>>>>>>>> On 12/02/2010 11:45 AM, Nicholas Butko wrote:
>>>>>>>>> In trying to compile pcl, I got an error from cminpack. The error reads
>>>>>>>>>
>>>>>>>>> echo "Building CMinpack..."
>>>>>>>>> Building CMinpack...
>>>>>>>>> cd build/cminpack-1.1.0&&     mkdir -p build&&     cd build&&     cmake -DUSE_FPIC=ON -DSHARED_LIBS=ON -DBUILD_EXAMPLES=OFF ..&&     make&&     cd ../../../../
>>>>>>>>> -- Operating system is generic Unix
>>>>>>>>> -- Building shared libraries.
>>>>>>>>> CMake Error at CMakeLists.txt:57 (install):
>>>>>>>>>  install TARGETS given no LIBRARY DESTINATION for shared library target
>>>>>>>>>  "cminpack".
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> -- Configuring incomplete, errors occurred!
>>>>>>>>> make: *** [installed] Error 1
>>>>>>>>> Granite:cminpack nick$
>>>>>>>>>
>>>>>>>>> The relevant line reads:
>>>>>>>>>
>>>>>>>>> install (TARGETS cminpack LIBRARY DESTINATION ${CMINPACK_LIB_INSTALL_DIR}
>>>>>>>>>    ARCHIVE DESTINATION ${CMINPACK_LIB_INSTALL_DIR} COMPONENT library)
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> Maybe CMINPACK_LIB_INSTALL_DIR needs to be set in the make script?
>>>>>>>>> _______________________________________________
>>>>>>>>> ros-users mailing list
>>>>>>>>> [hidden email]
>>>>>>>>> https://code.ros.org/mailman/listinfo/ros-users
>>>>>>>
>>>>>
>>>
>




_______________________________________________
[hidden email] / http://pointclouds.org
https://code.ros.org/mailman/listinfo/pcl-users
Reply | Threaded
Open this post in threaded view
|

Re: Error building pcl related to boost fusion

Nicholas Butko
I just tried to comment out that one line (POINT_CLOUD_REGISTER_WITH_FUSION in register_point_struct.h) and got a whole bunch of link errors mostly related to "at_key".

Patrick, can you do me a favor and check if my patch is doing "the right thing"? I am not very confident that it is, and I am getting a bunch of linking errors related to flann templated functions for the point types.  Things like 

{{{
  "pcl::KdTreeANN<pcl::PointXYZRGBA>::cleanup()", referenced from:
      pcl::RadiusOutlierRemoval<pcl::PointXYZRGBA>::applyFilter(pcl::PointCloud<pcl::PointXYZRGBA>&)in radius_outlier_removal.o
      pcl::KdTreeANN<pcl::PointXYZRGBA>::~KdTreeANN()in radius_outlier_removal.o
      pcl::KdTreeANN<pcl::PointXYZRGBA>::~KdTreeANN()in radius_outlier_removal.o
  "pcl::KdTreeANN<pcl::PointXYZRGBA>::radiusSearch(pcl::PointCloud<pcl::PointXYZRGBA> const&, int, double, std::vector<int, std::allocator<int> >&, std::vector<float, std::allocator<float> >&, int) const", referenced from:
      vtable for pcl::KdTreeANN<pcl::PointXYZRGBA>in radius_outlier_removal.o
  "pcl::KdTreeFLANN<pcl::PointXYZI>::nearestKSearch(pcl::PointXYZI const&, int, std::vector<int, std::allocator<int> >&, std::vector<float, std::allocator<float> >&)", referenced from:
      vtable for pcl::KdTreeFLANN<pcl::PointXYZI>in statistical_outlier_removal.o
}}}

--Nick 

On Dec 7, 2010, at 12:03 AM, Patrick Mihelich wrote:

Thanks for the report Nick. I should've known the Fusion stuff would bite me :P.

Radu, are we actually using the Fusion registration of point types in PCL at all? I put it in originally to dissociate struct fields from standardized channel names. So for example, you could have some arbitrary type

struct MyPointType
{
  float pos[3];
  // ...
};

And register pos[0], pos[1], pos[2] as channels "x", "y", "z". But if we're always assuming that point.x, point.y, point.z are valid, there's not much purpose.

Patrick

On Mon, Dec 6, 2010 at 11:27 PM, Nicholas Butko <[hidden email]> wrote:
This patch is simpler, doesn't use undocumented APIs, and does the same thing as the previous patch. However, I'm still not sure if it's doing the *right* thing, particularly because I am getting linking errors now. This is better than before, when I was getting compile errors.

I'm not sure if the linking errors are caused by this patch not doing the right thing, or are caused by something else entirely. This preprocessor stuff is pretty hard to debug, and I'd appreciate if someone with more experience than me could take a look.

--Nick





On Dec 6, 2010, at 10:14 PM, Nicholas Butko wrote:

> Radu,
>
> I finally had some time to sit down with this one, and I think I figured it out. However, someone much more familiar with boost than me should confirm this.
>
> It seems like "register_point_struct.h" uses a preprocessor macro that no longer exists in boost/fusion, specifically BOOST_FUSION_ADAPT_ASSOC_STRUCT_I. When I ran point_types.h through the preprocessor, this didn't get resolved. When I looked through my boost library (1.45), it didn't exist. When I looked through an older boost library (1.41) it did.
>
> So the next step was to confirm if it was relegated to macports or more general. I downloaded the boost source from boost.org and sure enough, the macro wasn't defined in the current version of boost. So it's not just a problem with macports.
>
> I am attaching a patch that I hope fixes the problem temporarily.  Unfortunately, it does this by calling undocumented macros like BOOST_FUSION_ADAPT_ASSOC_STRUCT_BASE and BOOST_FUSION_ADAPT_ASSOC_STRUCT_C. It would probably be best if register_point_struct.h could be rewritten to use only documented macros, like BOOST_FUSTION_ADAPT_ASSOC_STRUCT.
>
> I am not that confident in the patch, because basically I tried to guess what was going on from the implementation of BOOST_FUSTION_ADAPT_ASSOC_STRUCT. Someone should look it over to see if it's doing the right thing, and to make sure it doesn't break anything.
>
> --Nick
>
> <register_point_struct_h.patch>
>
>
> On Dec 3, 2010, at 1:03 PM, Radu Bogdan Rusu wrote:
>
>> CC-ing back the list, maybe some boost macos experts/users can help us identify this issue.
>>
>> So, the OpenMP thing should not be a big problem. If they're just warnings you can ignore them. In principle, the compiler should support this -- but I have no experience with MacOS.
>>
>> What about the Boost issues? Patrick do you happen to know where they might come from?
>>
>> Cheers,
>> Radu.
>>
>>
>> On 12/03/2010 12:37 PM, Nicholas Butko wrote:
>>> Jeez. Boost takes a long time to build.
>>>
>>> Anyway, the problem did not fix itself. My boost version is 1.45.0_1, my OS is OSX 10.6.5. The perception_pcl stack I am using is "unstable".
>>>
>>> If you have any questions or suggestions, I can try things out for you.
>>>
>>> By the way, in an unrelated issue, I also get lots of warnings related to openmp, such as
>>>
>>> -- Looking for omp_set_num_threads
>>> -- Looking for omp_set_num_threads - not found
>>> -- Looking for omp_set_num_threads
>>> -- Looking for omp_set_num_threads - not found
>>> -- Looking for omp_set_num_threads
>>> -- Looking for omp_set_num_threads - not found
>>> -- Looking for omp_set_num_threads
>>> -- Looking for omp_set_num_threads - not found
>>> -- Looking for omp_set_num_threads
>>> -- Looking for omp_set_num_threads - not found
>>> -- Looking for omp_set_num_threads
>>> -- Looking for omp_set_num_threads - not found
>>> WARNING: OpenMP compile flag not found
>>> WARNING: OpenMP compile flag not found
>>> WARNING: OpenMP compile flag not found
>>> WARNING: OpenMP compile flag not found
>>> WARNING: OpenMP compile flag not found
>>>
>>>
>>> On Dec 3, 2010, at 10:48 AM, Radu Bogdan Rusu wrote:
>>>
>>>> Whoops. Ok, now I really have to know what OS, inside/outside ROS.... :)
>>>>
>>>> Cheers,
>>>> Radu.
>>>>
>>>>
>>>> On 12/03/2010 10:45 AM, Nicholas Butko wrote:
>>>>> No problem.
>>>>>
>>>>> Unfortunately, now I'm having trouble building pcl. The problem seems to be with POINT_CLOUD_REGISTER_POINT_STRUCT(). I get errors like
>>>>>
>>>>> In file included from /Users/nick/ros/stacks/perception_pcl/pcl/include/pcl/range_image/range_image.h:45,
>>>>>                 from /Users/nick/ros/stacks/perception_pcl/pcl/src/pcl/range_image/range_image.cpp:46:
>>>>> /Users/nick/ros/stacks/perception_pcl/pcl/include/pcl/point_types.h:188: error: expected identifier before ‘(’ token
>>>>> /Users/nick/ros/stacks/perception_pcl/pcl/include/pcl/point_types.h:188: error: expected unqualified-id before ‘,’ token
>>>>> /Users/nick/ros/stacks/perception_pcl/pcl/include/pcl/point_types.h:188: error: expected `)' before ‘,’ token
>>>>> /Users/nick/ros/stacks/perception_pcl/pcl/include/pcl/point_types.h:188: error: expected `)' before ‘,’ token
>>>>> /Users/nick/ros/stacks/perception_pcl/pcl/include/pcl/point_types.h:188: error: ‘x’ has not been declared
>>>>> /Users/nick/ros/stacks/perception_pcl/pcl/include/pcl/point_types.h:188: error: expected constructor, destructor, or type conversion before ‘)’ token
>>>>>
>>>>> on every line with POINT_CLOUD_REGISTER_POINT_STRUCT. The POINT_CLOUD_REGISTER_POINT_WRAPPER does not seem to be a problem.
>>>>>
>>>>> --Nick
>>>>>
>>>>> On Dec 3, 2010, at 10:18 AM, Radu Bogdan Rusu wrote:
>>>>>
>>>>>> Thanks Nicholas! I pushed the patch in trunk.
>>>>>>
>>>>>> http://www.ros.org/wiki/perception_pcl/ChangeList
>>>>>>
>>>>>> Cheers,
>>>>>> Radu.
>>>>>>
>>>>>>
>>>>>> On 12/02/2010 03:35 PM, Nicholas Butko wrote:
>>>>>>> Radu,
>>>>>>>
>>>>>>> The attached patch works for me. The only change is to add -DCMINPACK_LIB_INSTALL_DIR="." to the cmake call.
>>>>>>>
>>>>>>> --Nick
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> On Dec 2, 2010, at 11:51 AM, Radu Bogdan Rusu wrote:
>>>>>>>
>>>>>>>> Nicholas,
>>>>>>>>
>>>>>>>> What version, OS, inside/outside ROS, etc? :)
>>>>>>>>
>>>>>>>> Cheers,
>>>>>>>> Radu.
>>>>>>>>
>>>>>>>>
>>>>>>>> On 12/02/2010 11:45 AM, Nicholas Butko wrote:
>>>>>>>>> In trying to compile pcl, I got an error from cminpack. The error reads
>>>>>>>>>
>>>>>>>>> echo "Building CMinpack..."
>>>>>>>>> Building CMinpack...
>>>>>>>>> cd build/cminpack-1.1.0&&     mkdir -p build&&     cd build&&     cmake -DUSE_FPIC=ON -DSHARED_LIBS=ON -DBUILD_EXAMPLES=OFF ..&&     make&&     cd ../../../../
>>>>>>>>> -- Operating system is generic Unix
>>>>>>>>> -- Building shared libraries.
>>>>>>>>> CMake Error at CMakeLists.txt:57 (install):
>>>>>>>>>  install TARGETS given no LIBRARY DESTINATION for shared library target
>>>>>>>>>  "cminpack".
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> -- Configuring incomplete, errors occurred!
>>>>>>>>> make: *** [installed] Error 1
>>>>>>>>> Granite:cminpack nick$
>>>>>>>>>
>>>>>>>>> The relevant line reads:
>>>>>>>>>
>>>>>>>>> install (TARGETS cminpack LIBRARY DESTINATION ${CMINPACK_LIB_INSTALL_DIR}
>>>>>>>>>    ARCHIVE DESTINATION ${CMINPACK_LIB_INSTALL_DIR} COMPONENT library)
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> Maybe CMINPACK_LIB_INSTALL_DIR needs to be set in the make script?
>>>>>>>>> _______________________________________________
>>>>>>>>> ros-users mailing list
>>>>>>>>> [hidden email]
>>>>>>>>> https://code.ros.org/mailman/listinfo/ros-users
>>>>>>>
>>>>>
>>>
>





_______________________________________________
[hidden email] / http://pointclouds.org
https://code.ros.org/mailman/listinfo/pcl-users