Environments for lb-conda

This repository contains the YAML files which define the software available from the lb-conda command. For information about using the environments see lhcb-core/lbcondawrappers.

Adding an environment

To add a new environment create a new file under the environments/ directory of this repository. Upon being merged the environment will be automatically deployed to CVMFS by GitLab CI.

What can be included in an environment

To be included packages should be available from conda-forge.

In the case of a Python package, the simplest first step is to make it pip-installable by add it to the Python Package Index (PyPI). For documentation about how to do this see the Python Packaging Guide. If you're distributing a Pure Python package (i.e. no compilled code) and unsure what to use, we recommend using setuptools with pyproject.toml. Additionally, consider adding the lhcbsoft as an owner of the package on PyPI so it's possible to keep ownership of the package even if people leave the collaboration.

Once the package is available with pip install PACKAGE_NAME, it can be added to conda-forge via staged-recipes. Typically grayskull can be used to automatically generate the conda-forge package recipe.

If this workflow appears to not be appropriate for your package please open an issue.

Updating an environment

Environments are never updated to help maintain reproducibility. Instead, each time one of the files under environments/ is modified, the given environment is created with a new "version" (corresponding to the current date and time). This way users of lb-conda can always switch back to a previous version. The modification to the environment YAML definition can either be to the definition of the environment or to simply add a comment.

If you are regularly updating an environment, feel free to request merge permissions on this repository.