README.md

Installing and Running the DECISION Conda Environment

The conda program is an open source package and environment manager commonly used for Python, and operating at the core of the Anaconda Python distribution. The YAML file(s) in this directory define a conda environment for running the DECISION python software.

Installing and Updating Anaconda

Instructions on installing Anaconda are here: https://docs.anaconda.com/anaconda/install/. Make sure Anaconda's bin directory is on your path before proceeding.

If Anaconda is already installed, but is not up to date, you can update it with the following commands (the first updates the conda program itself, the latter updates the python packages bundled with Anaconda):

$ conda update conda
$ conda update anaconda

Creating the Conda Environment

Assuming you have an up-to-date version of anaconda installed, you can install or update the environments with the commands below.

$ cd $DECISION_REPO
$ conda env create -f envs/environment.yml

In the above example, the variable $DECISION_REPO is assumed to point to the DECISION repo root.

Updating the Conda Environment

If the environment.yml file changes for any reason (e.g. adding new packages), the environment may be updated to be consistent with this new YAML file with the following command.

$ cd $DECISION_REPO
$ conda env update -f envs/environment.yml

In the above example, the variable $DECISION_REPO is assumed to point to the DECISION repo root.

Usage

With Anaconda's bin directory on your path, you can activate the environment by typing source activate decision (the environment's name, decision, is defined at the top of the YAML file). Once active, only the packages, and respective versions, installed into the environment will be available when running Python.

Exiting the environment is a simple source deactivate command at the command-line.

$ source activate decision
$ source deactivate

Exporting an Environment

Once created, an environment's definition may be exported to a YAML file again. The command to do this:

$ conda env export > environment.yml

This environment.yml file will contain all packages that were installed, which will be a superset on those originally provided, as it will include any dependencies needed by the packages specified in environment.yml. It will also contain the exact version of the installed packages, (hopefully) allowing an environment to be perfectly reproduced on another machine.

Adapted from ONWATCH enviornment documentation written by Steven Lewis.