rclcpp
rclcpp copied to clipboard
ros2
Add preempted ResultCode for WrappedResult and preempt API for ServerGoalHandle
The current version assumes the availability of ~~action_msgs::msg::GoalStatus::STATUS_ABORTED~~ action_msgs::msg::GoalStatus::STATUS_PREEMPTED. The code was written and tested with this addition in action_msgs. Does this call for a ticket requesting this addition in rcl_interfaces or should I make changes here?
I believe that rclpy action also needs to be updated based on this fix. Otherwise we lose the consistency.
If rclpy and action_msgs is also expected to make this update, it would make sense to add preemption to rcl_action altogether.
@naiveHobo do you mean it assumes the STATUS_PREEMPTED not STATUS_ABORTED in your description?
I don't necessarily think that we have to implement the python version as well. That could be left to another party that cares about that (which we don't) unless the maintainers here require it.
The cppcheck failures are happening on the master branch, so I guess it's not related: http://build.ros2.org/view/Fdev/job/Fdev__rclcpp__ubuntu_focal_amd64/25/testReport/
Waiting for https://github.com/ros2/rcl_interfaces/pull/105 to pass so action_msgs::msg::GoalStatus::STATUS_PREEMPTED can be used in this PR.
Thanks for linking that here. @jacobperron who's the owner of that one?
@jacobperron who's the owner of that one?
If you're referring to the rcl_interfaces PR, nobody yet; the core ROS 2 repositories have shared maintainership across the team. Typically, people will be assigned during triage meetings.
I can take a quick look at the proposed changes, but these kinds of API changes will have to wait until post-Foxy release.
Makes sense, I wouldn't expect otherwise. It would be nice to get in at a soon-after sync. This would be a critical fault for Navigation2 if this wasn't available in Foxy at all. It currently blocks the only reasonable solution some severe architectural issues.
