Using UCL JupyterHub

Creating an Environment (Staff)

1. Start up the UCL VPN.
2. Connect to JupyterHub
3. Authenticate using UCL credentials.
4. Create a new terminal: File > New > Terminal

Setting Up the Environment

The ISD instructions are not correct in the sense that the use of a symlink can cause problems and duplicated environments down the line. Anyway, type the following, but note that you need to replace {replace_this_with_group_name} with the appropriate group name (this will be obvious once logged in and you can run ls /shared/groups/):

course_name="casa0013"
group_name="{replace_this_with_group_name}"
conda config --add envs_dirs /shared/groups/${group_name}/casa
curl -o /tmp/${course_name}.yml https://raw.githubusercontent.com/jreades/sds_env/master/conda/environment_py.yml
conda env create -p /shared/groups/${group_name}/casa -f /tmp/${course_name}.yml

Tweaks to environyment_py.yml

Two shortcomings in the existing approach of generating environment_py.yml were identified and need to be tweaked in the Makefile:

1. Remove anything with linux in it
2. Remove version from gitpython.
3. Remove python-graphviz entirely.

Additional issues may exist with replication to non-Linux systems.