Example_13 test is broken

[christophfroehlich]

It seems that we broke example_13 (or at least the test) with ros-controls/ros2_control#2347

     [ros2_control_node-1] [INFO] [1753760884.651091427] [controller_manager]: Activating controllers: [ threedofbot_joint_state_broadcaster threedofbot_pid_gain_controller ]
    [ros2_control_node-1] [WARN] [1753760884.651350703] [controller_manager]: Component 'FakeThreeDofBot' is in INACTIVE state, but has start interfaces to switch: 
    [NON-XML-CHAR-0x1B][91mCould not switch controllers since prepare command mode switch was rejected.[NON-XML-CHAR-0x1B][0m
    [ros2_control_node-1] Start interfaces: 
    [ros2_control_node-1] [
    [ros2_control_node-1]   threedofbot_joint1/pid_gain
    [ros2_control_node-1]   threedofbot_joint2/pid_gain
    [ros2_control_node-1]   threedofbot_joint3/pid_gain
    [ros2_control_node-1] ]
    [ros2_control_node-1] Stop interfaces: 
    [ros2_control_node-1] [
    [ros2_control_node-1] ]
    [ros2_control_node-1] 
Error: s2_control_node-1] [ERROR] [1753760884.651373715] [controller_manager]: Could not switch controllers since prepare command mode switch was rejected.

#859
#861
#862
#863
ros-controls/ros2_control_ci#421
ros-controls/ros2_control_ci#422
ros-controls/ros2_control_ci#423
ros-controls/ros2_control_ci#424
ros-controls/ros2_control_ci#425

[christophfroehlich]

@saikishor I analyzed this failure. The demo starts the threedofbot_pid_gain_controller while the fake bot (generic system) is in inactive state.

Configure FakeThreeDofBot and its joint state broadcaster and non-movement command interfaces. Call

.. code-block:: shell

ros2 control set_hardware_component_state FakeThreeDofBot inactive
ros2 control switch_controllers --activate threedofbot_joint_state_broadcaster threedofbot_pid_gain_controller

Your PR now throws an error here, even before the prepare_command_mode_switch method of the hardware is called.

Is it possible to have command interfaces (like PID gains in this example) if the hardware is inactive?

[saikishor]

@christophfroehlich right now it doesn't let you activate any interfaces if it is in an inactive state. We will have to implement the logic of differentiating interfaces soon.

[christophfroehlich]

neither GPIO command is allowed in inactive state of the HW?

[christophfroehlich]

Isn't this then wrong now?
https://github.com/ros-controls/ros2_control/blob/68e190ac5a4447150a7ea0a2457d945a8ab228d8/hardware_interface/include/hardware_interface/actuator_interface.hpp#L69-L74