packing-packages 0.4.0

This module provides functionality to pack conda environments and their dependencies into a specified directory.

PACKING-PACKAGES

Overview

Packing-Packages is a powerful tool for packaging conda environments and their dependencies into portable directories. Perfect for deploying Python environments to offline machines, air-gapped systems, or ensuring reproducible environments across different machines with the same operating system.

Key Features

• 🚀 Simple & Fast: Pack entire conda environments with a single command
• 📦 Complete Dependency Resolution: Automatically includes all conda and pip dependencies
• 🔄 Cross-Platform Support: Pack from YAML files even on different operating systems
• 🛠️ Flexible Installation: Install packages directly or generate standalone install scripts
• ⚡ Incremental Updates: Use --diff-only to download only new packages
• 🎯 Error Resilient: Installation continues even if some packages fail, with detailed error reporting
• 📋 YAML-Based Packing: Create packages directly from conda environment YAML files

Use Cases

• Deploy Python environments to offline or air-gapped systems
• Create reproducible environment snapshots for production deployments
• Migrate conda environments between machines without internet access
• Archive complete environment states for backup or compliance
• Prepare environment packages for different target platforms

Installation

pip install packing-packages

Requirements: Python, setuptools, and pip

Quick Start

1. Pack Your Environment

On your source machine (with internet access), pack the current conda environment:

conda activate myenv
packing-packages pack -d .

This creates a directory structure:

.
└── myenv/
    ├── conda/    # Conda packages
    └── pypi/     # PyPI packages

2. Install on Target Machine

On your destination machine (offline), install the packed packages:

conda create -yn myenv --offline
conda activate myenv
packing-packages install ./myenv

Alternative: Generate Standalone Scripts

Generate install scripts that work independently without requiring packing-packages:

packing-packages install ./myenv --generate-scripts

This creates install_packages.bat (Windows) and install_packages.sh (Unix/Linux) that can be executed directly on the target machine.

Advanced Features

Pack from YAML Files

Pack environments directly from conda environment YAML files, even on a different operating system:

packing-packages pack yaml environment.yaml -p linux-64

Incremental Updates

Only download packages that don't already exist in the target directory:

packing-packages pack -d . --diff-only

Error-Resilient Installation

The packing-packages install command continues even if some packages fail, providing detailed error reports for manual handling.

Documentation

For detailed command-line options, advanced usage, and troubleshooting, see the full documentation or the command-line tutorial.

Important Notes

Installation Methods

You can install packages using either:

• Standard commands (conda install / pip install): Faster, but stops on first error
• packing-packages install: Slower, but continues on errors and provides detailed reports

Special Cases

• Non-PyPI sources (e.g., PyTorch official index): May require manual handling
• Older packages: Some may need --use-pep517 flag for successful installation
• Proxy environments: Set HTTP_PROXY and HTTPS_PROXY environment variables

For detailed troubleshooting and edge cases, please refer to the command-line tutorial.