Environment management is crucial to reproducibility of work and for making your work runnable in hosted containers such as through Binder or Collab. This folder should be used for keeping track of your dependencies. We STRONGLY advise that you use some kind of dependency manager and not just pip install stuff locally!
If you would prefer to work with something other than conda environments, be sure to check out this page on what kind of files can be used to configure a Binder environment. If you choose to go with a conda environment, then the only file that needs to be edited is conda/environment.yml. However if you use anything else some extra setup is going to be required. This is because Binder checks the root folder and the binder folder for configuration files. The binder folder already contains a symlink pointing to conda/environment.yml, but if you choose to use say, conda/requirements.txt instead you will need a symlink pointing to that file from the binder folder instead.
To create a symlink on windows, use the command mklink, of the form mklink {link_path} {relative_path}. As an example, the following command is how the existing binder symlink was created:
mklink binder\environment.yml ..\conda\environment.yml
Although we refer to "conda" environments, we recommend using mamba as a drop in replacement for the conda package manager. Mamba performs operations in parallel, which we've found to be faster for creating complex environments involving many packages!
The only file you should need to edit in this folder is conda/environment.yml. This file defines the set of conda-packages needed to render the full website.
conda env create --name geosmart --file=environment.yml
conda activate geosmart
conda env update --file environment.yml --prune
Lockfiles ensure that everyone working on this project has an identical development environment, whether working on a personal computer or running on our hosted JupyterHub cloud infrastructure.
If you edit conda/environment.yml to change package versions or add new ones, be sure to re-lock the environment by running ./lock-environment.sh:
mamba remove --name geosmart --all
mamba env create --name geosmart --file conda-linux-64.lock.yml
mamba activate geosmart