Add utility function to get datetime.datetime from Time

[tonynajjar]

Feature request

Feature description

I found myself needing to convert from rclpy's Time to datetime.datetime to input that in a third-party library. Would it make sense to add such a utility function? If so I could open a PR

Implementation considerations

I think it would just look like that:

return datetime.datetime.fromtimestamp(self._time_handle.nanoseconds / 1e9)

[gordon-n-stevenson]

This is an excellent idea, as I think it is a common issue when trying to timestamp out things and the management of the seconds/nanoseconds into a correct datetime can be challenging for new users.

It wouldn't be far off something like this SO answer - https://stackoverflow.com/questions/15649942/how-to-convert-epoch-time-with-nanoseconds-to-human-readable - but with the use of rclpy.Time right?

[JeanPaulSB]

Hi everyone, I'm a new contributor, can I work on this as my first issue? thx

[sloretz]

@JeanPaulSB Welcome!

I think you could take this as a first issue. However, time is always more complicated than one would think it should be. There might be something subtle we're not thinking of. I recommend reading the ROS Contributing guide, but here's some more targeted guidance:

• Target the rolling branch with your PR
• Add a new function or method to do the conversion. I would recommend Time.to_datetime() to match the existing API Time.to_msg()
• Add a new test to test_time.py

Something subtle about time in ROS is we support different kinds of Clocks. ROS_TIME and SYSTEM_TIME both make sense to me to be convertible to a datetime object, but STEADY_TIME (or UNINITIALIZED) does not. I recommend making Time.to_datetime() raise an exception if the time object's clock_type is not one of ClockType.ROS_TIME or ClockType.SYSTEM_TIME.

Feel free to ask me questions here as you go. I'm happy to help!

[JeanPaulSB]

@sloretz Thank you so much for guidelines before addressing it, I'll be working on it and I'll be in touch with you if there are any questions, thanks again for the help.

[JeanPaulSB]

@sloretz Hi! hope you are doing well. I'm trying to build rclpy from source with my forked repo, however, it is throwing me this error when building with colcon build --symlink-install

Starting >>> rclpy   
--- stderr: rclpy                             
/app/rclpy/rclpy/src/rclpy/action_server.cpp: In member function ‘void rclpy::ActionServer::configure_introspection(rclpy::Clock&, pybind11::object, rcl_service_introspection_state_t)’:
/app/rclpy/rclpy/src/rclpy/action_server.cpp:380:19: error: ‘rcl_action_server_configure_action_introspection’ was not declared in this scope; did you mean ‘rcl_service_configure_service_introspection’?
  380 |   rcl_ret_t ret = rcl_action_server_configure_action_introspection(
      |                   ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
      |                   rcl_service_configure_service_introspection
/app/rclpy/rclpy/src/rclpy/action_client.cpp: In member function ‘void rclpy::ActionClient::configure_introspection(rclpy::Clock&, pybind11::object, rcl_service_introspection_state_t)’:
/app/rclpy/rclpy/src/rclpy/action_client.cpp:282:19: error: ‘rcl_action_client_configure_action_introspection’ was not declared in this scope; did you mean ‘rcl_client_configure_service_introspection’?
  282 |   rcl_ret_t ret = rcl_action_client_configure_action_introspection(
      |                   ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
      |                   rcl_client_configure_service_introspection
gmake[2]: *** [CMakeFiles/_rclpy_pybind11.dir/build.make:132: CMakeFiles/_rclpy_pybind11.dir/src/rclpy/action_server.cpp.o] Error 1
gmake[2]: *** Waiting for unfinished jobs....
gmake[2]: *** [CMakeFiles/_rclpy_pybind11.dir/build.make:104: CMakeFiles/_rclpy_pybind11.dir/src/rclpy/action_client.cpp.o] Error 1
gmake[1]: *** [CMakeFiles/Makefile2:212: CMakeFiles/_rclpy_pybind11.dir/all] Error 2
gmake[1]: *** Waiting for unfinished jobs....
gmake: *** [Makefile:146: all] Error 2
---

I'm using Ubuntu 24.04 noble.

Should I build ROS2 from source to make it work? Thank you so much for your help.

[christophebedard]

Yeah, we sometimes make changes to one repository/package that requires changes in another package. In this case, you're building rclpy @ rolling from source (I assume), and it uses a recent change to rcl (ros2/rcl#1207). If you just build all of ROS 2 from source, it should work. (Otherwise, you can try to figure out which exact packages need to be built from source, but that's a bit annoying and it can change at any time.)

If you do use binaries, do make sure you're using the Rolling binaries and not the Jazzy binaries, though.

[JeanPaulSB]

@christophebedard I was mixing binaries and building other packages from source but It ended up being a nightmare because I had to figure out which ones need to be built from source as you pointed out so I finally moved on by building everything from source. Thanks for your response. I have already raised the implementation on #1443.