Add new get_value API for Handles and Interfaces

[saikishor]

Related to ros-controls/ros2_controllers#1442 and ros-controls/ros2_controllers#1443

There are still some deprecated warnings in the following ros2_control packages:

transmission_interface: Once this PR is merged. I will perform the changes here : #1845
hardware components: This needs to be handled in a different PR.

I'll take care of both

[codecov]

Codecov Report

Attention: Patch coverage is 87.69231% with 16 lines in your changes missing coverage. Please review.

Project coverage is 89.36%. Comparing base (f8c03cc) to head (45d8e71).

Files with missing lines	Patch %	Lines
...de/hardware_interface/loaned_command_interface.hpp	41.66%	6 Missing and 1 partial ⚠️
...lude/hardware_interface/loaned_state_interface.hpp	41.66%	6 Missing and 1 partial ⚠️
...re_interface/include/hardware_interface/handle.hpp	66.66%	1 Missing and 1 partial ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #1976      +/-   ##
==========================================
+ Coverage   89.35%   89.36%   +0.01%     
==========================================
  Files         130      130              
  Lines       14576    14570       -6     
  Branches     1258     1254       -4     
==========================================
- Hits        13024    13021       -3     
+ Misses       1088     1086       -2     
+ Partials      464      463       -1     

Flag	Coverage Δ
unittests	89.36% <87.69%> (+0.01%)	⬆️

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

Files with missing lines	Coverage Δ
...nclude/semantic_components/force_torque_sensor.hpp	97.95% <100.00%> (ø)
...terface/include/semantic_components/imu_sensor.hpp	97.77% <100.00%> (+0.10%)	⬆️
...erface/include/semantic_components/pose_sensor.hpp	96.55% <100.00%> (ø)
...mantic_components/semantic_component_interface.hpp	90.90% <100.00%> (ø)
...rface/test/test_chainable_controller_interface.cpp	100.00% <100.00%> (ø)
...chainable_controller/test_chainable_controller.cpp	88.37% <100.00%> (+0.27%)	⬆️
...r_manager/test/test_controller/test_controller.cpp	88.73% <100.00%> (ø)
...t_controllers_chaining_with_controller_manager.cpp	99.25% <100.00%> (+0.02%)	⬆️
...rface/test/mock_components/test_generic_system.cpp	99.72% <ø> (-0.04%)	⬇️
...dware_interface/test/test_component_interfaces.cpp	96.49% <100.00%> (+0.15%)	⬆️
... and 7 more

... and 2 files with indirect coverage changes

[christophfroehlich]

Thanks for pushing this forward.

I was thinking of something shorter than 2xget+2xvalue in get().get_value<double>().value(), but still have no better suggestion.

[christophfroehlich]

One thing: Could you please update the release_notes and also add something to the migration notes? We have only a section about the hardware_components there.

[saikishor]

One thing: Could you please update the release_notes and also add something to the migration notes? We have only a section about the hardware_components there.

Sure I'll do that

[destogl]

It is not beautiful, but probably we don't have a better way to do this…

Maybe we think of a simpler API in the future. At least we are good with the types.

Maybe the approach from ROS 2 parameters using method get_value_as_double() is better compared to get_value<double>().value()?

@bmagyar what do you think?

[destogl]

Do we actually need this format of get_value? Are we using this somewhere of do you have specific use-case where this has advantages over std::optional<T> get_value()?

[saikishor]

Right now, we are not using it in ros2_control repositories, but this exists from many releases both in rolling nd jazzy. So, at least not to break API of people that might use it, I haven't removed it here

[destogl]

Also this version, do we need it?

[saikishor]

Same here as well

[saikishor]

Maybe the approach from ROS 2 parameters using method get_value_as_double() is better compared to get_value<double>().value()?

If we use optional, we cannot avoid using .has_value() and .value() even with the new method as the ROS2 parameters. That's a bit downside.

I have used .value() directly here as we know most of these tests work synchronously, so we always have value, but in the ROS2 Controllers etc, we will need to have the .get_value() as usual into a variable and check if it has value and if it does have value, then replace other instances of the variable in the code with .value()