6.3. Python Install Modes🔗

The Python package (python/l4yaml) is pure Python over ctypes, so it has no compile step — only an install step. The CMake driver supports three install paradigms, picked by the L4YAML_PYTHON_INSTALL option:

ament — uses ament_cmake_python from a sourced ROS 2 environment. Installs the package at ${prefix}/lib/pythonX.Y/site-packages/l4yaml/; colcon's setup.bash automatically prepends that path to PYTHONPATH. This is the right mode for a ROS 2 / colcon workflow.
venv — runs pip install -e python from ${L4YAML_PYTHON_VENV}/bin/python at configure time. The install is editable, so source edits in python/l4yaml/ take effect without re-running CMake.
none — CMake does not touch Python; the user installs the package manually with pip install python or equivalent.

The default auto mode picks ament when ament_cmake_python is on the CMake prefix path (i.e. a ROS underlay is sourced), venv when L4YAML_PYTHON_VENV is set, otherwise none. The same colcon build command therefore works in both a ROS shell (uses ament) and a plain dev shell (falls back to none), without flag changes.

6.3.1. ROS 2 / colcon🔗

In a colcon workspace with a ROS underlay sourced, no additional flags are needed:

source /opt/ros/jazzy/setup.bash
colcon build --packages-select L4YAML

After build, source the workspace overlay (. install/setup.bash) and import l4yaml works.

6.3.2. venv (non-ROS)🔗

For local development outside ROS, point the CMake driver at a venv:

python -m venv .venv && . .venv/bin/activate
cmake -B build -S . \
      -DL4YAML_PYTHON_INSTALL=venv \
      -DL4YAML_PYTHON_VENV=$VIRTUAL_ENV
cmake --build build -j

The pip install -e python runs at configure time and the venv keeps tracking python/l4yaml/ thereafter.