Frequently Asked Questions

Frequently Asked Questions#

What are conda-forge and miniforge?#

Conda-forge is a community maintained repository of conda packages hosted on the conda-forge GitHub organisation. Conda-forge is built on top of the popular Anaconda package manager but it doesn’t use the default Anaconda package repository, which has some license restrictions for commercial use.

Miniforge is the preferred conda-forge installer[1], which will install the conda package, and environment, manager and setup conda-forge.

Why do CAF recommend conda-forge and Miniforge?#

  • Conda-forge is a community maintained repository, which allows CAF packages to be released on it and means it generally has more packages available than Anconda’s default repository.

  • Conda-forge doesn’t have the license restricting commercial use, which Anaconda’s default package repository does have.

  • The conda package manager is very popular, well used and, with new updates, fast at creating environments.

There are other Python package managers available and one which was considered was poetry. Poetry provides some additional benefits vs conda although these were not deemed significant enough to outweigh the much greater popularity of conda based systems.

Why do CAF recommend requirements.txt files?#

Python has a few different formats for defining dependencies, the main four are:

  • requirements.txt: the most basic file format, which can be used by most tools.

  • environment.yml: a format specific to conda.

  • Pipfile: a format specific to pipenv.

  • pyproject.toml: a general Python configuration file which provides different sections for many development and build tools.

The main benefits of requirements.txt are:

  • Not specific to a single package manager allowing users to use their preferred system.

  • Much simpler to define and more flexible on the allowed values, e.g. allows defining packages to be installed from URLs or Git repositories.

  • Works with CAF’s package build method setuptools, which enables installation of CAF packages with pip or conda.

See also

Requirements File section of Environments for more details and examples.

Contents