Setup¶

We assume a working Python installation as described in Requirements. pybrid can be installed either from pre-built binaries served on PyPI, which is the right path for most users, or from source for developers and on platforms without published wheels.

In both cases, we recommend installing pybrid into a dedicated virtual environment to avoid dependency conflicts with the rest of your Python setup. While you can manage virtual environments manually, we find the uv package manager to be the most convenient option, and for the remainder of this documentation we assume that uv is used for environment and package management.

Installing from pre-built binaries¶

Both packages that make up pybrid (see Requirements) are available as pre-built wheels on PyPI for the supported operating systems and architectures. Setting up a fresh environment and installing pybrid with uv takes two commands:

# create a new virtual environment in the .venv/ folder
# using Python 3.13 (the recommended version)
uv venv --python 3.13

# install pybrid-computing (pulls in pybrid-computing-native automatically)
uv pip install pybrid-computing

To update an existing installation to the latest release (something you should do from time to time), pass the -U flag:

uv pip install -U pybrid-computing

Keep both package versions in sync

The version numbers of pybrid-computing and pybrid-computing-native must always be identical. Installing or updating pybrid-computing via uv pip install keeps the two aligned automatically, but if you later pin or upgrade one of them by hand, double-check that the versions still match.

Installing from source¶

Installing from source is the right path for users on a platform without pre-built wheels, or for anyone who wants to hack on pybrid itself. It additionally requires a working C++ compiler supporting at least the C++14 standard and cmake (see Requirements for the full list), and the initial build typically takes several minutes because the native extension is compiled locally.

After checking out the source, run the following two commands from the repository root:

uv venv --python 3.13

# build both packages and install them into the venv
uv sync

To produce distributable binary wheels (for example to share them with other machines of the same platform), invoke uv build once per package:

uv build packages/pybrid-computing-native
uv build packages/pybrid-computing

This places the platform-specific .whl files under the respective dist/ directories. Updating a source installation is a matter of pulling the latest state from git and re-running uv sync to rebuild both packages.