Also initialize non-joint components

[fmauch]

Before, only joints were reading the initial value parameter. However, also sensors and gpios can have configured initial values.

Initialization of vectors were called, but the implementation only checked for joints. This leaves all sensor state_interfaces initialized with NaN, which in turn breaks other things such as UniversalRobots/Universal_Robots_ROS2_Driver#390.

With the changes suggested here, also sensors and gpios get initialized with the initial values read from the URDF.

I think in conjunction it would make sense to extend the ros2_control_demos with inital values. I've used those as a simplified environment also to verify that my URDF was not the culprit here.

I am not 100% sure whether this would be the most suitable modification, as I haven't completely understood the dataflow here, but that change seemed rather obvious to me. Feel free to point me into another direction if needed.

Anyway, this should also be ported back to Galactic. Probably a direct backmerge is not possible, since the fake->mock renaming didn't happen there, but it should be easy to do nonetheless.

[destogl]

Thanks! This is great!

Excellent idea with the example in ros2_control_demos.

Could you also add a test for this in this file? It should be straightforward. One should add another xml example and then check if parsing is good. There is already test for other initial values. Extending that test would also be fine.

[fmauch]

Could you also add a test for this in this file? It should be straightforward. One should add another xml example and then check if parsing is good.

sure!

[fmauch]

Oh, I was not aware, that by re-requesting a review I would remove other potential reviewers. Sorry if I destroyed the workflow there.

[destogl]

Oh, I was not aware, that by re-requesting a review I would remove other potential reviewers. Sorry if I destroyed the workflow there.

No this shouldn't be happening. It's not your fault!

[destogl]

FYI: I just converted it to draft and back to re-trigger reviewer-lottery workflow. (closing was unintentional :D)

[destogl]

This looks great! Thanks!

[destogl]

Thanks for following up!

[codecov-commenter]

Codecov Report

Merging #822 (32a67bb) into master (925f5f3) will decrease coverage by 2.04%.
The diff coverage is 38.09%.

@@            Coverage Diff             @@
##           master     #822      +/-   ##
==========================================
- Coverage   34.61%   32.57%   -2.05%     
==========================================
  Files          52       91      +39     
  Lines        2981     9295    +6314     
  Branches     1855     6251    +4396     
==========================================
+ Hits         1032     3028    +1996     
- Misses        310      720     +410     
- Partials     1639     5547    +3908     

Flag	Coverage Δ
unittests	32.57% <38.09%> (-2.05%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
controller_manager/src/controller_manager.cpp	36.63% <ø> (-3.08%)	⬇️
controller_manager/src/realtime.cpp	0.00% <0.00%> (ø)
controller_manager/src/ros2_control_node.cpp	0.00% <0.00%> (ø)
..._interface/include/hardware_interface/actuator.hpp	100.00% <ø> (ø)
...re_interface/include/hardware_interface/sensor.hpp	100.00% <ø> (ø)
...re_interface/include/hardware_interface/system.hpp	100.00% <ø> (ø)
hardware_interface/src/system.cpp	55.45% <ø> (ø)
...rface/test/mock_components/test_generic_system.cpp	9.14% <ø> (ø)
...dware_interface/test/test_component_interfaces.cpp	32.44% <ø> (+4.25%)	⬆️
hardware_interface/test/test_component_parser.cpp	8.52% <ø> (-3.13%)	⬇️
... and 107 more

[fmauch]

Any updates on merging this? The failing tests seem to be unrelated but I might be wrong there.

[destogl]

@Mergifyio backport humble

[mergify]

backport humble

✅ Backports have been created

• #991 Also initialize non-joint components (backport #822) has been created for branch humble