Packaging

This section explains how the build process works and how to manually build wheels. It should give insights into how package is deployed on PyPI and how the plugin system is setup.

Default build process

The FastEMRIWaveforms package relies on scikit-build-core to handle the build process. This tool makes use of CMake to handle the compilation of C/C++/CUDA backends while being compatible with modern python packaging practices (i.e. by using a pyproject.toml file instead of the legacy setup.py approach).

When installing few from sources, or simply building a wheel package, the main steps executed are:

  • Create a temporary isolated environment

  • Install in that environment the build system dependencies listed in pyproject.toml: scikit-build-core, numpy, cython, setuptools_scm, …

  • Let scikit-build-core orchestrate the next steps:

    • Read the main CMakeLists.txt file to detect the required CMake version

    • Detect whether a corresponding cmake command is available, otherwise automatically install it in the current isolated environment

    • Call setuptools_scm to detect the current project version from git tags and write it to src/few/_version.py

    • Call CMake to handle backends compilation

    • Package the resulting compiled modules and the project Python sources into a wheel

Building core and plugin packages

By default, CMake will always compile at least the CPU backend, and will try to also compile the GPU backend if required dependencies are available. This results in a single wheel which contains (1) the pure-python core package, (2) the compiled CPU backend and optionally (3) a compiled GPU CUDA backend.

This is ideal for local development and installation from source, but different from how FEW is released on PyPI where 3 packages are deployed:

  • fastemriwaveforms: contains the python code and the CPU backend

  • fastemriwaveforms-cuda11x: contains only the cuda11x GPU backend

  • fastemriwaveforms-cuda12x: contains only the cuda12x GPU backend

The process for building these differenciated wheels is defined in .github/workflows/publish.yml which handles this build process and package deployment to PyPI. The core logic is handled by tweaking slightly the pyproject.toml file before building each category of wheels.

Common steps

Some steps are performed for all built wheels:

# Change the version scheme to force a clean version like "1.5.2" instead of 1.5.2.post1.dev51+gfe23bf1.d20250218
sed -i 's|version_scheme = "no-guess-dev"|version_scheme = "only-version"|g' pyproject.toml
sed -i 's|local_scheme = "node-and-date"|local_scheme = "no-local-version"|g' pyproject.toml

Building the core package

To build the core package wheel, the following command is executed after the common steps for python 3.9 to 3.13:

pip wheel ./ --no-deps -w ./dist \
  --config-settings=cmake.define.FEW_WITH_GPU=OFF

The wheels built on Linux are, by default, distribution specific and must be made into manylinux wheels to improve their compatibility with many distributions. This is done by repairing the wheels using auditwheel:

for whl in ./dist/*.whl; do
    auditwheel repair "${whl}" -w ./wheelhouse/ --plat manylinux_2_27_x86_64
done

The manylinux wheels will be put into ./wheelhouse.

Building the GPU plugin packages

To build GPU plugin packages, multiple modifications to pyproject.toml must be applied:

# Change the project name to add the `-cuda11x` or `-cuda12x` suffix
sed -i 's|" #@NAMESUFFIX@|-cuda12x"|g' pyproject.toml

# Add `cupy-cuda11x` or `cupy-cuda12x` to the project dependencies
sed -i 's|#@DEPS_CUPYCUDA@|"cupy-cuda12x"|g' pyproject.toml

# Add a dependency of the project core package
sed -i 's|#@DEPS_FEWCORE@|"fastemriwaveforms"|g' pyproject.toml

# Delete the line containing the falg @SKIP_PLUGIN@ from pyproject.toml
# that line instruct scikit-build-core to add the directory src/few to the wheel
# so deleting it removes all python sources from the generated wheel
sed -i '/@SKIP_PLUGIN@/d' pyproject.toml

The wheels are then built for python 3.9 to 3.13 with the command:

pip wheel ./ --no-deps -w ./dist \
  --config-settings=cmake.define.FEW_WITH_GPU=ONLY

The option FEW_WITH_GPU=ONLY instructs CMake to build a GPU backend and to skip the CPU one. therefore, in the end, the wheel contains only the compiled modules for the GPU backend.

Just like for the core package, the wheels must be repaired to become manylinux wheels. Since they have a dependencies on NVIDIA dynamic libraries, they are not strictly-speaking manylinux compatible but mechanisms are in place on the core package Python code to detect issues with these dependencies and advise the user about required steps. auditwheel must be instructed to ignore those external dependencies like so:

for whl in ./dist/*.whl; do
    auditwheel repair "${whl}" -w /wheelhouse/ \
        --plat manylinux_2_27_x86_64 \
        --exclude "libcudart.so.12" \
        --exclude "libcusparse.so.12" \
        --exclude "libcublas.so.12" \
        --exclude "libnvJitLink.so.12" \
        --exclude "libcublasLt.so.12"
done
# Replace the .so.12 extension by .so.11 if you are building the cuda11x plugin

Understanding the CMake compilation mechanism

CMake is a powerful scripting language used to manage the compilation steps of complex projects. One of its main advantage is its cross-platform compatibility: it provides abstraction layers to make the compilation independent from the current operating system and thus makes the building steps of FEW working on both Linux, macOS, and Windows (though that last OS has not been tested thoroughly).

CMake is unfortunately also known having convoluted syntaxes and for having many examples that make use of outdated/legacy syntaxes. This is why, for FEW, most of CMake complexity is “hidden” from developers.

When declaring a compiled module with CMake with a declaration like

few_add_lib_to_cuda_cpu_backends(
  NAME pyinterp
  PYX_SOURCES pyinterp.pyx
  CU_SOURCES interpolate.cu CPU_LINK PUBLIC "lapacke;lapack"
  HEADERS cuda_complex.hpp global.h matmul.hh
  INCLUDE PRIVATE ${Python_NumPy_INCLUDE_DIR})

The following steps are executed:

  • CMake calls the following command to process the Cython file pyinterp.pyx into a C++ file:

$ python -m cython pyinterp.pyx \
    --output-file pyinterp.cpp \
    -3 \ # Select Python 3 syntax
    -+ \  # Build C++ output instead of C
    --module-name pyinterp \
    -I ./  # Search for header files in local directory

  • If the CPU backend needs to be built, declare that a dynamic library few_backend_cpu/pyinterp.cpython-3X-${arch}.so must be built with:

    • C++ source files (or interpreted as such): pyinterp.cpp and interpolate.cu

    • Header files: cuda_complex.hpp, global.h and matmul.hh

    • Linking to the libraries referenced by the CMake targets lapacke and lapack (equivalent to the compilation flags -L/path/to/lapack/lib -llapacke -llapack)

  • If a GPU backend needs to be build, declare that a dynamic library few_backend_cuda12x/pyinterp.cpython-3X-${arch}.so must be built with:

    • C++ source files: `pyinterp.cpp``

    • CUDA source file (compiled with nvcc): interpolate.cu

    • The same headers files that for the CPU backend

    • Link to CUDA libraries (-lcudart -lcublas -lcusparse)

    • Specific compilation flags related to device linking, CUDA architecture, …

All that machinery is implemented in cmake/FEW.cmake and actually makes uses of filename manipulation, temporary directories, file copying to differentiate compiling the same file in C++ mode and in CUDA mode, etc… But these are implementation details, only elements declared in the call to few_add_lib_to_cuda_cpu_backends should be of importance as detailed in the dedicated section.