Python Package Tools in 30 mins

TL;DR: Comprehensive guide to Python package managers and virtual environment tools, from pip and venv to modern alternatives like Poetry and uv.

Summary

Python's ecosystem offers numerous tools for managing packages and virtual environments. Choosing the right tool depends on:

• Project requirements: Simple scripts vs. complex applications
• Dependency management: Basic installs vs. sophisticated resolution
• Environment isolation: Simple venv vs. cross-language support
• Performance needs: Standard tools vs. fast Rust-based alternatives
• Publishing needs: Development-only vs. PyPI distribution

This guide organizes tools by category (built-in, modern, specialized) to help you select the best fit for your workflow.

Built-In and Standard Tools

pip

What it does: Default Python package installer that downloads and installs packages from PyPI

Key capabilities:

• Installs packages from PyPI

• Works with requirements.txt for dependency specifications

  # Install a single package from PyPI
  > pip install requests
  
  # Install all packages listed in requirements.txt (common for project setup)
  > pip install -r requirements.txt
  
  # Export all installed packages with their versions to requirements.txt (for sharing/reproducing environment)
  > pip freeze > requirements.txt
  
  # Install a specific version of a package (for version pinning)
  > pip install requests==2.28.0
  
  # Upgrade a package to the latest version
  > pip install --upgrade requests
  
  # Uninstall a package
  > pip uninstall requests
  

• Best for: Basic package installation and simple projects

venv

What it does: Creates isolated Python environments to separate project dependencies and avoid conflicts

Key capabilities:

• Built-in virtual environment tool (Python 3.3+)

• Isolated environments with their own Python interpreter and packages

  # Create a new virtual environment named 'venv' in the current directory
  > python -m venv venv
  
  # Activate the virtual environment on macOS/Linux (makes Python and pip point to the venv)
  > source venv/bin/activate
  
  # Activate on Windows
  > venv\Scripts\activate.bat
  
  # Deactivate the current virtual environment (returns to system Python)
  > deactivate
  
  # Create a venv without pip (if you want to install pip separately)
  > python -m venv venv --without-pip
  
  # Remove a virtual environment (just delete the directory)
  > rm -rf venv
  

• Best for: Basic environment isolation without extra dependencies

virtualenv

What it does: Creates isolated Python environments with more features and flexibility than venv

Key capabilities:

• Older alternative to venv but still maintained and enhanced
• Python version discovery and environment templates

• Slightly faster environment creation

  # Install virtualenv globally (required before first use)
  > pip install virtualenv
  
  # Create a new virtual environment named 'myenv'
  > virtualenv myenv
  
  # Create a virtualenv with a specific Python version (if multiple Pythons are installed)
  > virtualenv -p python3.11 myenv
  
  # Create a virtualenv using system site-packages (inherits globally installed packages)
  > virtualenv --system-site-packages myenv
  
  # Activate the virtualenv (same as venv)
  > source myenv/bin/activate
  
  # List available Python interpreters that virtualenv can use
  > virtualenv --discovery cached
  

• Best for: Advanced virtual environment needs or pre-Python 3.3 projects

Modern Dependency and Environment Managers

pipenv

What it does: Unifies package management and virtual environment creation into a single tool

Key capabilities:

• Combines pip + venv functionality
• Uses Pipfile and Pipfile.lock for deterministic dependency resolution

• Automatic virtual environment management

  # Install a package and automatically create/update Pipfile and Pipfile.lock
  > pipenv install requests
  
  # Activate a subshell within the virtual environment (creates venv if it doesn't exist)
  > pipenv shell
  
  # Install dev dependencies (e.g., testing tools, linters)
  > pipenv install --dev pytest
  
  # Install all dependencies from Pipfile.lock (for reproducible deployments)
  > pipenv sync
  
  # Update all packages to their latest compatible versions
  > pipenv update
  
  # Run a command within the virtual environment without activating shell
  > pipenv run python script.py
  

• Best for: Simple project workflows requiring both dependency and environment management

Poetry

What it does: Comprehensive tool for dependency management, packaging, and publishing using modern standards

Key capabilities:

• Modern dependency management with sophisticated resolution
• Uses pyproject.toml (PEP 518 standard)
• Built-in dependency resolution and lockfile generation

• Handles packaging and publishing to PyPI

  # Initialize a new Poetry project interactively (creates pyproject.toml)
  > poetry init
  
  # Add a package as a dependency (updates pyproject.toml and poetry.lock)
  > poetry add requests
  
  # Activate a shell within the virtual environment
  > poetry shell
  
  # Add a dev dependency (e.g., testing or linting tools)
  > poetry add --group dev pytest
  
  # Install all dependencies from poetry.lock (for setting up project on new machine)
  > poetry install
  
  # Build distribution packages (wheel and sdist) for publishing
  > poetry build
  
  # Publish package to PyPI
  > poetry publish
  
  # Update dependencies to their latest compatible versions
  > poetry update
  
  # Run a command within the virtual environment without activating shell
  > poetry run python script.py
  

• Best for: Modern Python applications and libraries requiring robust dependency management and publishing

Conda

What it does: Cross-language package and environment manager handling Python packages and system dependencies

Key capabilities:

• Cross-language package + environment manager
• Popular in data science and scientific computing communities
• Manages Python versions and binary dependencies

• Installs pre-compiled binaries for faster installation

  # Create a new environment with a specific Python version
  > conda create -n myenv python=3.11
  
  # Activate an environment (makes it the active Python)
  > conda activate myenv
  
  # Install a package from conda channels (pre-compiled binaries)
  > conda install numpy
  
  # Install multiple packages at once (recommended for better dependency resolution)
  > conda install numpy pandas matplotlib
  
  # Export environment to a file (for sharing/reproducing)
  > conda env export > environment.yml
  
  # Create environment from a YAML file
  > conda env create -f environment.yml
  
  # Deactivate current environment (returns to base)
  > conda deactivate
  
  # List all environments
  > conda env list
  
  # Remove an environment
  > conda env remove -n myenv
  

• Best for: Data science and scientific computing projects requiring complex binary dependencies

Mamba

What it does: Drop-in replacement for Conda with faster C++ dependency solver

Key capabilities:

• Faster alternative to conda with compatible commands and packages
• Faster dependency solver written in C++

• Particularly beneficial for large, complex environments

  # Create a new environment (same as conda but much faster)
  > mamba create -n myenv python=3.11
  
  # Activate environment (use conda activate, not mamba activate)
  > conda activate myenv
  
  # Install packages with faster dependency resolution than conda
  > mamba install numpy pandas scikit-learn
  
  # Update packages (significantly faster than conda update)
  > mamba update --all
  
  # Search for available packages
  > mamba search tensorflow
  
  # Install from a YAML file (faster than conda)
  > mamba env create -f environment.yml
  

• Best for: Large scientific environments where conda's solver is too slow

uv

What it does: Ultra-fast Rust-based package installer and resolver with 10-100x performance improvement

Key capabilities:

• Drop-in pip replacement with compatible CLI
• Extremely fast Rust-based package manager
• Can manage virtual environments

• Supports lockfiles for reproducible installs

  # Create a virtual environment (10-100x faster than venv)
  > uv venv
  
  # Install packages (significantly faster than pip)
  > uv pip install requests
  
  # Install from requirements.txt (with parallel downloads)
  > uv pip install -r requirements.txt
  
  # Compile dependencies to a lockfile (for reproducible installs)
  > uv pip compile requirements.in -o requirements.txt
  
  # Sync environment to exactly match lockfile (removes unlisted packages)
  > uv pip sync requirements.txt
  
  # Install a specific version with fast resolution
  > uv pip install "django>=4.0,<5.0"
  

• Best for: Fast modern workflows where performance matters, especially for CI/CD pipelines

pipx

What it does: Installs and runs Python CLI applications in isolated environments without dependency conflicts

Key capabilities:

• Installs Python CLI tools globally in isolated environments
• Each tool gets its own virtual environment
• Tools are globally accessible as commands

• Prevents dependency conflicts between CLI tools

  # Install a CLI tool globally in its own isolated environment
  > pipx install black
  
  # Install another tool (gets separate venv, no dependency conflicts)
  > pipx install pytest
  
  # Run a tool temporarily without installing it (downloads, runs, then cleans up)
  > pipx run cowsay "Hello!"
  
  # Upgrade an installed tool to the latest version
  > pipx upgrade black
  
  # List all installed tools and their versions
  > pipx list
  
  # Uninstall a tool and its isolated environment
  > pipx uninstall black
  
  # Upgrade all installed tools at once
  > pipx upgrade-all
  

• Best for: Installing and managing CLI tools like black, pytest, or cookiecutter globally

Comparison Overview

Tool	Manages Packages	Manages Venv	Lockfile	Python Version Mgmt	Best For
pip	Yes	No	No	No	Basic installs
venv	No	Yes	No	No	Environment isolation only
virtualenv	No	Yes	No	No	Legacy / advanced isolation
pipenv	Yes	Yes	Yes	No	Simple project workflows
poetry	Yes	Yes	Yes	No	Modern apps and libraries
conda	Yes	Yes	No	Yes	Data science
mamba	Yes	Yes	No	Yes	Large scientific environments
uv	Yes	Yes	Yes	No	Fast modern workflows
pipx	Yes	Auto	No	No	Installing CLI tools