sample-python-project

Accompanying code for the article "Python packaging for developers in a hurry".

The toolset:

• mamba (from the mambaforge distribution) for managing Python environments and installing packages which are only available in conda's repositories
• pip-tools for managing and pinning direct and transitive dependencies for applications (libraries are only loosely-pinned)
• setuptools for packaging the project with a declarative configuration in setup.cfg
• build and twine for sane and foolproof packaging and publishing of libraries on PyPI
• docker for packaging code and runtime environment of applications

The benefits:

• Effortless and consistent workflow across Linux, Windows and macOS
• Access to a vast catalog of packages that are hard-to-install, available from conda's repositories (mamba)
• Fast, easy-to-manage, and human-readable generation of pinned application dependencies for reproducible environments (pip-tools)
• Simple, easy-to-manage declarative configuration for packaging the project (setuptools and setup.cfg)
• Easy to get right, hard to get wrong building of source and wheel distributions for libraries (build and twine)
• Fast-to-build and cache-friendly docker images for applications, with a helpful dev stage for usage in CI environments (docker)

Contents

• application: contains a sample project structure for an application (where pinning dependencies, simple builds and isolated environments are key)
• library: contains a sample project structure for a library (where maximizing compatibility, and robust and reproducible building of source and wheel distributions are important)

Each folder contains a README.md that guides you through the (very similar) steps for getting started with them.