ament_cmake VS ros-noetic

This is the definition for ament_get_recursive_properties(). As you can see, it takes only 2 positional arguments, but an invisible 3rd "ARGN" seems to be passed, so even I am confused. As you said, _ would be a placeholder meaning var_libraries would not be set at the end of the function and would be skipped. It's hard to unravel the functions, so I might just message(STATUS "") the variables to see what is being passed and where. Any help pointing me in the right direction of solving the segmentation fault would be much appreciated!

Check out https://robostack.github.io/. It has distributions of ROS which can be installed on a PinePhone.

Robostack (https://robostack.github.io)

Take a look at robo stack - they distribute ROS through condaforge - https://robostack.github.io/

One could use https://robostack.github.io/ to install ROS tools inside a python environment isolated from the rest of the machine

The Robostack project is packaging ROS core distributions and many third-party ROS packages built against existing conda-forge dependencies as much as possible to make things more convenient for roboticists. That's mostly how I've been using ROS for well over a year now. The virtual env isolation is great, because otherwise robotics code tends to get tied to a specific Linux OS version. Disk space is cheap, I'd rather have the flexibility and a few 4GB environments lying around.