-
Notifications
You must be signed in to change notification settings - Fork 545
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Jazzy CI and build status #2851
Conversation
building jazzy docker images. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Note that the tutorial images fail to build:
For humble, due to a broken generate_parameter_library
:
/root/ws_moveit/build/moveit_kinematics/kdl_kinematics_plugin/kdl_kinematics_parameters/include/kdl_kinematics_parameters.hpp:146:36: error: expected unqualified-id before ‘.’ token
2522
146 | auto& entry = updated_params..joints_map[value_1];
For rolling due to missing package gazebo_ros2_control
:
kortex_bringup: Cannot locate rosdep definition for [gazebo_ros2_control]
.github/workflows/ci.yaml
Outdated
- IMAGE: jazzy-ci-testing | ||
ROS_DISTRO: jazzy |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Note that ci
and ci-testing
are identical since #2793: Because osrf/ros2:testing
(the base image of ci-testing
) was broken upstream at the time, I disabled building it and aliased it to ci
.
- IMAGE: jazzy-ci-testing | |
ROS_DISTRO: jazzy |
I had a look into the open issues in MTC's CI - based on rolling docker images. I observed strange segfaults during shutdown of binaries. It turned out, that two ABI-incompatible versions of If that proposal is not accepted, at least To test my hypothesis that octomap was the culprit, I manually created a Tests confirm that the segfaults are fixed: the number of failing tests reduced from 18 to 4 (actually only 2).
Interestingly, Humble does not exhibit that issue. I'm afraid that's a regression in Rolling's rclcpp. No idea, where this originates from. And I need to turn to different tasks next... So, this stays unresolved for now. Any help is welcome ;-) |
Once this succeeds, please merge with REBASE so that we keep the commit history. There are commits like dac7638 that should be reverted later on. |
Build all succeed now, we still get a bunch of segfaults in the unit test. In particular there is a double free in PILZ when testing under noble
|
I have fixed that in MoveIt1 and just pushed the corresponding port. |
4dfa3d7
to
889c112
Compare
Avoid double-free of KDL::RotationalInterpolation in KDL >= 1.4.1 The underlying memory leak was fixed in KDL with orocos/orocos_kinematics_dynamics#180
Unfortunately, you force-pushed and thus overwrote my commit for the Pilz double-free... |
Yeah, I just noticed this too late... Thank you, this helps a lot! |
I don't think 48f60c4 had anything to do with the double free, unfortunately. The same issue has been fixed a while ago in #1229 already. Oddly, the tests produce segfaults (in CI and also locally) with the compile guard, and since there is no active distro that still provides kdl <1.4.0 I'll remove this switch. All pilz tests pass locally for me without any issues on clang+jazzy with both of the two latest commits. However, the main issue left is that CI will still install the ros-octomap package, regardless if it's provided in the docker image or not. I don't think there is a way around a proper version guard in the cmake setup for geometric_shapes and moveit |
adf04b8
to
14bc36e
Compare
This reverts commit 55df0bc. The deprecated constructor was being used in the same file for the exact use case of enabling namespaces that are not specified by the parameter. There is no replacement for supporting a dynamic server lookup, however the parameter logic could still use simplification.
rolling with clang-tidy should be happy now as well (there were several new fixes). |
Thank you! I was planning to do a fresh cherry-pick of all the things that we keep, and squash it into concise blocks in a fresh branch. |
pulling in henningkayser/lcov-action
I'm switching off ccov for now (#2866) and will reopen this PR with a new branch to unblock CI and releases. |
No description provided.