conda-forge

It is a community led collection of recipes, build infrastructure, and packages for the conda package manager.

The problem we are trying to solve

$ conda search my-awesome-pkg

PackagesNotFoundError: The following packages are not available from current channels:

  - my-awesome-pkg

We want to avoid this scenario!

conda-skeleton to the rescue

conda skeleton pypi my-awesome-pkg

or

conda skeleton cran my-awesome-pkg

To use 'conda skeleton', install conda-build.

The recipe

{% set version = "1.0.0" %}

package:
  name: my-awesome-pkg
  version: {{ version }}

source:
  url: https://URL/my-awesome-pkg-{{ version }}.tar.gz
  sha256: 410bcd1d6409026fbaa65d9ed33bf6dd8b1e94a499e32168acf

The recipe

build:
  number: 0
  noarch: python
  script: python -m pip install --no-deps --ignore-installed .

requirements:
  build:
    - python
    - pip
  run:
    - python

The recipe

test:
  imports:
    - my-awesome-pkg
  commands:
    - my-awesome-pkg-cli --help

The recipe

about:
  home: http://my-awesome-pkg.org/
  license: Public Domain
  summary: 'Dummy package'
  license_family: OTHER

extra:
  recipe-maintainers:
    - ocefpaf

Build the recipe locally or with CIs

• docker pull condaforge/linux-anvil
• conda build my-awesome-pkg
• enable Travis-CI and AppVeyor on your fork

Submit the recipes to conda-forge

• check if the recipe was created the latest version of conda-skeleton
• check conda-forge’s recipe example/instructions
• submit a PR to https://github.com/conda-forge/staged-recipes

What happens next…

• the recipe will be built for OS X (Travis-CI), Linux (CircleCI), and Windows (AppVeyor)
• the community volunteers will review and merge the recipe
• once merged the recipe will be removed from staged-recipes
• a new repository, feedstock, will be created and the recipe maintainers will have write access to it

Maintaining a package

• the maintainer is not responsible updating, reviewing and merging PRs to the feedstock
• some of those PRs will be automatic updates to conda-forge’s boilerplate CI configuration
• in the near-future PRs updating the package will also be automatic, reducing the burden on the maintainer.

Hands on exercise!

• install conda-build
• choose a package from CRAN or PyPI
• run the conda-skeleton command for the package
• “clean” the recipe for conda-forge
• submit a PR to staged-recipes