Jacob Perron

If we decide to pass allocators individually, it'd be nice if the `rmw` allocator is API-compatible with the `rcl` allocator. Then we can just forward the allocators that users provide...

I don't know what the advantages are in typedef-ing the allocator; `rcl` does this. Maybe it allows us to change our minds later and define a custom allocator without changing...

I agree that we might want to change the logic to return False (instead of raising an error). The current logic does not seem very Pythonic.

The change you suggest sounds reasonable to me. I'm guessing it is simply a bug and not intended logic. I'm happy to review a PR; please target the default branch...

Could you provide an example of how you are declaring parameters in a launch file? Also, what are you trying to achieve by doing this? I would avoid calling `rclpy`...

The test appears flaky; retriggering the Rpr job resulting in all tests passing.

@dzywater Do you know if `rqt` worked as expected on Windows in previous releases? Based on this ticket, I don't think rqt_gui_cpp is expected to be working on Windows: https://github.com/ros-visualization/rqt/issues/178....

@hidmic No, I didn't look into it too deeply.

The error is referring to a service called `SetCloudLayerState`, which is not in your CMakeLists.txt. Is the CMakeLists.txt you've posted up-to-date?

@anacsousa1 Since this issue is labeled as "backlog", I take it no one is currently looking at it (at least no one from the ROS 2 team).