Python-cpp-experiments

My experiments with C++ extensions for python using pybind11.

 

Prerequisites

• Python 3.8–3.10 with pip (version 22+ recommended for editable installs)
• CMake (>= 3.16)
• Armadillo C++ linear algebra library — install via your package manager:
  • macOS: brew install armadillo
  • Ubuntu/Debian: sudo apt install libarmadillo-dev
• A C++17 compiler (e.g. AppleClang, GCC, MSVC)

Build

Option 1: scikit-build-core (recommended)

1. Initialize git submodules (required for carma):

   git submodule update --init --recursive

2. Install the project in editable mode:

   pip install -e ".[dev]"

Option 2: Manual cmake build

1. Initialize git submodules (required for carma):

   git submodule update --init --recursive

2. Install Python build dependencies:

   pip install "pybind11[global]>=2.6.0" "numpy>=1.14"

3. Build with cmake:

   mkdir build
   cd build
   cmake .. -DPYTHON_EXECUTABLE=$(python3 -c "import sys; print(sys.executable)") -DCMAKE_INSTALL_PREFIX:PATH=..
   make install

   Note: -DCMAKE_INSTALL_PREFIX:PATH is set to .. to make the path in the install command in the CMakeLists.txt file relative to the location of the file.

TODOs:

• use scikit-build-core to compile the C++ modules
• CI pipelines for tests
• Update Makefile
• remove sphinx docs (won't ever publish them for this project)
• Check how to automate the Armadillo installation either as a github submodule or using cmake's FetchContent like in carma