You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
In file disturbance_manager.py, single-threaded executor is used to process callbacks of two callback groups: default callback group of the node and reentrant callback group used for rate objects. 'rclpy' rate object internally implements timer callback and assigns those callbacks to the reentrant callback group. Therefore, this code has 1) service client objects and callbacks of the timer object assigned to the default mutually exclusive callback group and 2) callbacks of the rate object assigned to the separate reentrant callback group. As discussed in ROS documentation, callbacks from reentrant callback group can execute in parallel as well as callbacks from two callback groups can execute in parallel. However due to the single thread of the executor, the callbacks can not execute in parallel. Using a multi-threaded executor would be more efficient for this use case.
I would appreciate to hear your thoughts on this.
The text was updated successfully, but these errors were encountered:
In file disturbance_manager.py, single-threaded executor is used to process callbacks of two callback groups: default callback group of the node and reentrant callback group used for rate objects. 'rclpy' rate object internally implements timer callback and assigns those callbacks to the reentrant callback group. Therefore, this code has 1) service client objects and callbacks of the timer object assigned to the default mutually exclusive callback group and 2) callbacks of the rate object assigned to the separate reentrant callback group. As discussed in ROS documentation, callbacks from reentrant callback group can execute in parallel as well as callbacks from two callback groups can execute in parallel. However due to the single thread of the executor, the callbacks can not execute in parallel. Using a multi-threaded executor would be more efficient for this use case.
I would appreciate to hear your thoughts on this.
The text was updated successfully, but these errors were encountered: