conda-forge and Miniforge¶

conda-forge is a community-driven collection of conda packages. Miniforge is a minimal installer that uses conda-forge by default — free for all uses, including commercial.

Why conda-forge?¶

Anaconda Licensing Issue¶

Since 2020, Anaconda Inc. requires a paid license for commercial use (organizations with 200+ employees) when using: - Anaconda Distribution - Miniconda (with default channel) - The defaults channel

The Solution: conda-forge¶

conda-forge is: - Community-maintained (not owned by Anaconda Inc.) - Free for all uses (personal and commercial) - Has more packages than defaults channel - Often has newer package versions

What is conda-forge?¶

A channel (package repository) with: - 20,000+ packages - Community-maintained recipes - CI/CD for all platforms (Linux, macOS, Windows) - Transparent governance

Website: https://conda-forge.org

Miniforge vs Miniconda¶

Feature	Miniconda	Miniforge
Maintained by	Anaconda Inc.	conda-forge community
Default channel	`defaults`	`conda-forge`
Commercial use	License required	✅ Free
Solver	conda (slow)	conda or mamba

Miniforge = Miniconda but defaults to conda-forge channel.

Installing Miniforge¶

Download¶

From: https://github.com/conda-forge/miniforge

# Linux (x86_64)
wget https://github.com/conda-forge/miniforge/releases/latest/download/Miniforge3-Linux-x86_64.sh
bash Miniforge3-Linux-x86_64.sh

# macOS (Intel)
wget https://github.com/conda-forge/miniforge/releases/latest/download/Miniforge3-MacOSX-x86_64.sh
bash Miniforge3-MacOSX-x86_64.sh

# macOS (Apple Silicon)
wget https://github.com/conda-forge/miniforge/releases/latest/download/Miniforge3-MacOSX-arm64.sh
bash Miniforge3-MacOSX-arm64.sh

# Windows
# Download .exe from GitHub releases page

Verify¶

conda --version
conda config --show channels
# Should show: conda-forge (not defaults)

Mambaforge¶

Mambaforge = Miniforge + mamba pre-installed.

Mamba is a faster drop-in replacement for conda (see mamba).

# Linux
wget https://github.com/conda-forge/miniforge/releases/latest/download/Mambaforge-Linux-x86_64.sh
bash Mambaforge-Linux-x86_64.sh

# Then use mamba instead of conda
mamba install numpy pandas

Switching Existing Miniconda to conda-forge¶

If you already have Miniconda/Anaconda:

Option 1: Change Default Channel¶

# Remove defaults, add conda-forge
conda config --remove channels defaults
conda config --add channels conda-forge
conda config --set channel_priority strict

Option 2: Fresh Install (Recommended)¶

Export your environments:
```
conda env export -n myenv > myenv.yml
```
Uninstall Miniconda/Anaconda
Install Miniforge
Recreate environments:
```
conda env create -f myenv.yml
```

Using conda-forge Channel¶

Install from conda-forge¶

# Explicit channel
conda install -c conda-forge numpy

# If conda-forge is default
conda install numpy

Verify Package Source¶

conda list numpy

Output shows channel:

# Name                    Version                   Build  Channel
numpy                     1.24.0          py311h8e6699e_0    conda-forge

environment.yml with conda-forge¶

name: myproject
channels:
  - conda-forge
  # Don't include 'defaults' for license-free usage
dependencies:
  - python=3.11
  - numpy
  - pandas
  - scikit-learn
  - matplotlib
  - jupyter
  - pip
  - pip:
    - some-pip-package

Create:

conda env create -f environment.yml

conda-forge vs defaults¶

Aspect	defaults	conda-forge
Maintainer	Anaconda Inc.	Community
License	Commercial restrictions	Free
Package count	~8,000	~20,000+
Update frequency	Slower	Faster
Quality control	Anaconda team	Community + CI
ARM support	Limited	Better

Best Practices¶

1. Use Miniforge for New Installations¶

# Fresh start with conda-forge
bash Miniforge3-Linux-x86_64.sh

2. Set Strict Channel Priority¶

conda config --set channel_priority strict

Prevents mixing packages from different channels.

3. Don't Mix defaults and conda-forge¶

# Good
channels:
  - conda-forge

# Avoid
channels:
  - conda-forge
  - defaults    # Can cause conflicts

4. Pin Channels in environment.yml¶

Always specify channels in your environment file for reproducibility.

Troubleshooting¶

Package Not Found¶

# Check if package exists on conda-forge
conda search -c conda-forge package_name

# Or use pip as fallback
pip install package_name

Slow Solving¶

Use mamba instead of conda:

mamba install numpy pandas scikit-learn

Channel Conflicts¶

# Check current channels
conda config --show channels

# Reset to conda-forge only
conda config --remove channels defaults
conda config --add channels conda-forge

Summary¶

Distribution	Default Channel	License	Recommended
Anaconda	defaults	Commercial restrictions	❌
Miniconda	defaults	Commercial restrictions	❌
Miniforge	conda-forge	✅ Free	✅
Mambaforge	conda-forge	✅ Free	✅✅

Key Takeaways¶

conda-forge is community-maintained, free for commercial use
Miniforge is the recommended installer (conda-forge default)
Mambaforge = Miniforge + mamba (faster)
Avoid defaults channel for commercial projects
Set channel_priority: strict to prevent mixing
Most packages are available on conda-forge