URDF not parsable from python wrapper when bio_ik is find_package'd #29
Comments
timonegk
added a commit
to bit-bots/bitbots_motion
that referenced
this issue
Feb 19, 2021
timonegk
added a commit
to bit-bots/bitbots_motion
that referenced
this issue
Feb 19, 2021
6 tasks
|
I can't reproduce this on a custom non-debian build of ROS either.
This is missing Did you look through the linked libraries of bio_ik and the system dependencies for this? I would expect that bio_ik (obviously compiled in your own workspace) might link to a different version of some system library on your melodic system, resulting in ABI breakage and symbol clashes. |
Yes, you're right, but the error occurs before moveit complains about the missing semantics. It is also not a problem with my setup, I can reproduce the issue in a docker container. I attached an archive containing the reproducible example: minimal_example.tar.gz Extract the archive, navigate to the folder, run |
|
It is also not a problem with my setup, I can reproduce the issue in a docker container.
I attached an archive containing the reproducible example: [minimal_example.tar.gz](https://github.com/TAMS-Group/bio_ik/files/6021725/minimal_example.tar.gz)
Thanks for debugging this. I'm not saying it's a problem with your setup.
It might as well be a problem with the build of a ROS package.
Thank you for setting up the docker image. I'm not sure I'll have the time to hunt this down though. :(
Could you paste the output of `ldd -v` for all relevant binaries (the python module, the bio_ik object, the robot model loader) and is there any discrepancy there?
|
ldd for python module with bio_ikOutput of ldd for python module without bio_ik |
Output for libbio_ik.soOutput for libmoveit_robot_model_loader.so |
timonegk
added a commit
to bit-bots/bitbots_motion
that referenced
this issue
Feb 25, 2021
PeterMitrano
pushed a commit
to UM-ARM-Lab/arm_robots
that referenced
this issue
May 14, 2021
currently breaks loading the robot description when used from python!? see TAMS-Group/bio_ik#29
|
👍 I also have this problem when using python bindings that use bio_ik |
|
👍 I also have this problem when using python bindings that use bio_ik
Feel free to look into this. I planned to investigate it, but did not find the time to do that until now. :sad:
|
|
So these lines in the CMakeLists.txt seems suspicious, removing them fixes the problem for me! Why link to these statically? |
|
Removing the lines fixes the problem for me, too. |
PeterMitrano
pushed a commit
to UM-ARM-Lab/arm_robots
that referenced
this issue
Jul 10, 2021
also fixed bug (?) in bio_ik linking TAMS-Group/bio_ik#29
PeterMitrano
pushed a commit
to UM-ARM-Lab/arm_robots
that referenced
this issue
Jul 13, 2021
also fixed bug (?) in bio_ik linking TAMS-Group/bio_ik#29
|
Thank you for finding the cause of the issue.
```
-static-libgcc
-static-libstdc++
```
Why link to these statically?
Probably the same reasons why you would link any library statically.
Did you benchmark whether shared linking lowers performance?
Even if there should be a slight effect, fixing the issue is probably worth it from my side.
|
|
Hm I didn't know static linking affect runtime performance -- I did not benchmark anything. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
The problem
I am using the MoveIt RobotModelLoader to load a URDF from the ros parameter server. The URDF will not be parsed when the following two conditions are fulfilled:
CMakeLists.txtReproducible example:
cd catkin_ws/src && catkin create pkg minimal_exampleCMakeLists.txt
wrapper.cpp
roslaunch pr2-description upload_pr2.launchThen, a lot of ros errors are raised:
Now, remove
bio_ikfrom theCMakeLists.txt, recompile, and try again. The problem will not occur.I tested this on ROS Melodic on Ubuntu 18.04 for both python2 and python3 (just replace python with python3 in the
find_package(BOOST...)). The problem did not occur on a self compiled workspace on Debian Testing.In my case it was fortunately possible to remove
bio_ikfrom the catkin components, but that may not always be the case.I'm sorry for the bizarre issue.
The text was updated successfully, but these errors were encountered: