packing-packages 0.3.1rc0

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

PACKING-PACKAGES

This tool packs a conda environment and all its dependencies into a directory.

You can migrate the packed environment to another offline machine with the same operating system.

Install

# Need python, setuptools, pip package
pip install packing-packages

How to use

See help for available commands and options:

packing-packages --help
packing-packages pack --help
packing-packages install --help

Pack

packing-packages pack -d .

Install

packing-packages install .

Generate install scripts (instead of installing directly)

Use --generate-scripts to create reusable install scripts that do not depend on this package. These scripts can be copied and executed on another machine (offline), and work across platforms (Windows batch and Unix/Linux shell).

# Generate scripts in the package directory (default)
packing-packages install . --generate-scripts

# Specify environment name and output directory explicitly
packing-packages install /path/to/packages \
  --generate-scripts \
  --env-name myenv \
  --output-dir /path/to/output

Generated files:

• install_packages.bat (Windows)
• install_packages.sh (Unix/Linux)

Notes:

• If --env-name is omitted with --generate-scripts, the environment name defaults to the package directory name.
• If --output-dir is omitted, scripts are written to the package directory.

Example

Source device

conda activate <envname>
python -m pip install packing-packages
python -m packing_packages pack -d .

Destination device (offline)

conda create -yn <envname> --offline
conda activate <envname>
conda install --use-local --offline ./conda/*
python -m pip install --no-deps --no-build-isolation ./pypi/*

Alternatively, if you generated install scripts on the source device:

# Windows
install_packages.bat

# Unix/Linux
./install_packages.sh

Notes

Installing PyTorch from a Non-PyPI Source

If you need to download packages from a source other than PyPI (e.g., the official PyTorch index), standard methods may not complete successfully. Some packages may fail to download and require manual handling. Automated support is not currently provided for that case.

# Installation command
pip install torch==2.5.1 --index-url https://download.pytorch.org/whl/cu124

# Download command (without dependencies)
# --no-deps: download only the specified package
# -d: set the destination directory
pip download torch==2.5.1 --index-url https://download.pytorch.org/whl/cu124 --no-deps -d .

Handling Installation Failures with --use-pep517

Some older packages may not install successfully using pip install. In such cases, installing with the --use-pep517 option may help:

pip install <package-name> --use-pep517

Choosing Between Standard Install Commands and packing-packages install

There are two main methods for installing packages:

1. Using conda install or pip install
2. Using the packing-packages install command, which wraps these tools and provides error handling

Command	Advantages	Disadvantages
conda install / pip install	Fast execution	Stops immediately upon error
packing-packages install	Skips failed packages and reports them	Slower overall process

Packing an Environment Using a .yaml File

If you already have an environment file (.yaml), you can create and pack the environment on an online machine with the same OS:

You can now pack from a YAML file even on a different OS using the following command:

packing-packages pack yaml /path/to/file.yaml

Using a Proxy (If Required)

If you are in an environment that requires a proxy, you may need to configure the proxy settings before downloading or installing packages:

export HTTP_PROXY="your proxy"
export HTTPS_PROXY="your proxy"