Python Packaging Tools: A Complete Guide
Python's packaging ecosystem can feel overwhelming at first — there are tools for installing packages, creating virtual environments, building distributions, and publishing to PyPI. This guide breaks down the most important tools every Python developer should know, what each one does, and when to use it.
1. pip — The Package Installer
pip is Python's default package installer. It comes bundled with Python and is used to install, upgrade, and remove packages from the Python Package Index (PyPI).
pip install requests pip install requests==2.31.0 pip uninstall requests pip list pip freeze > requirements.txt pip install -r requirements.txt
Use it for: Installing packages in almost any Python project. It's the foundation nearly every other tool builds on.
2. venv — Built-in Virtual Environments
venv is the standard library module for creating isolated Python environments, so project dependencies don't clash with each other or with system packages.
python -m venv myenv source myenv/bin/activate # Linux/Mac myenv\Scripts\activate # Windows deactivate
Use it for: Simple, dependency-free environment isolation. No installation required since it ships with Python.
3. virtualenv
virtualenv is the older, more feature-rich predecessor to venv. It supports older Python versions and offers a few extra options venv doesn't.
pip install virtualenv virtualenv myenv
Use it for: Legacy projects or when you need features venv lacks (e.g., faster environment creation, Python 2 support).
4. setuptools
setuptools is the classic library for defining how a Python project is packaged, historically configured via setup.py or setup.cfg.
from setuptools import setup, find_packages
setup(
name="mypackage",
version="0.1.0",
packages=find_packages(),
install_requires=["requests"],
)
Use it for: Building and distributing packages, especially in older or existing codebases still using setup.py.
5. wheel
wheel is a built-package format (.whl) that installs faster than the older source-distribution format because it skips the build step at install time.
pip install wheel python setup.py bdist_wheel
Use it for: Producing distributable, pre-built packages for faster installs.
6. build
build is the modern, standards-based tool (PEP 517/518) for building both source distributions (sdist) and wheels from a pyproject.toml file.
pip install build python -m build
Use it for: The current recommended way to build distributable packages, replacing setup.py build.
7. twine
twine is used to securely upload your built packages to PyPI or a private package index.
pip install twine twine upload dist/*
Use it for: Publishing your package to PyPI after building it with build or setuptools.
8. Poetry
Poetry is an all-in-one dependency management and packaging tool. It handles virtual environments, dependency resolution, building, and publishing — all through a single pyproject.toml file.
curl -sSL https://install.python-poetry.org | python3 - poetry new myproject poetry add requests poetry install poetry build poetry publish
Use it for: Modern projects that want a single tool for dependency management, virtual environments, and publishing, with reliable lockfile-based reproducibility.
9. Pipenv
Pipenv combines pip and virtualenv into one workflow, using a Pipfile and Pipfile.lock instead of requirements.txt.
pip install pipenv pipenv install requests pipenv shell pipenv lock
Use it for: Application development (not library packaging) where you want reproducible environments with a lockfile.
10. conda
conda is a language-agnostic package and environment manager, popular in the data science and scientific computing communities because it can install non-Python dependencies too (e.g., compiled C libraries).
conda create -n myenv python=3.11 conda activate myenv conda install numpy pandas
Use it for: Data science and machine learning projects that depend on complex binary or non-Python dependencies.
11. Hatch
Hatch is a newer, modern project manager that handles environments, builds, versioning, and publishing, aiming to be a lightweight alternative to Poetry with strong plugin support.
pip install hatch hatch new myproject hatch build hatch publish
Use it for: Projects wanting a flexible, standards-compliant tool with built-in environment and version management.
12. PDM
PDM is another modern package manager built around PEP 582 and pyproject.toml, offering fast dependency resolution and optional support for running without virtual environments.
pip install pdm pdm init pdm add requests pdm install
Use it for: Developers wanting fast resolution and flexible environment handling with modern standards.
Quick Comparison
| Tool | Main Purpose | Best For |
|---|---|---|
| pip | Install packages | Everyone, always |
| venv | Environment isolation | Built-in, simple projects |
| setuptools | Build packages | Classic packaging |
| build | Build sdist/wheel | Modern standards-based builds |
| twine | Publish to PyPI | Uploading packages |
| Poetry | Dependency + packaging | Modern all-in-one workflow |
| Pipenv | Dependency + env management | Application development |
| conda | Cross-language packages | Data science / ML |
| Hatch | Project management | Lightweight modern workflow |
| PDM | Fast dependency resolution | Modern PEP 582/pyproject projects |
Final Thoughts
If you're just starting out, pip + venv is all you need. As your projects grow — especially if you plan to publish packages or need reproducible builds — tools like Poetry, Hatch, or PDM streamline the entire workflow into a single command-line tool. For data science work involving non-Python dependencies, conda remains the go-to choice.
Understanding these tools and how they fit together will save you countless hours of dependency headaches down the road.
0 Comments