The Zeus election server

LICENCE: This code is released under the GPL v3 or later

This is a fork of Ben Adida's Helios server. The differences from Helios are as follows:

• Whereas Helios produces election results, Zeus produces a tally of the ballots cast.

• This allows Zeus to be used in voting systems other than approval voting (which is supported by Helios), since the vote tally can be fed to any other system that actually produces the election results.

• In terms of overall architecture and implementation it is closer to the original Helios implementation than Helios v. 3.

Install

Install Postgres (postgres-server, libpq-dev).

Install the following libraries:

libgmp-dev libmpfr-dev libmpc-dev

If necessary, create a Postgres user. Then create a database:

sudo -u postgres createuser -s $(whoami)
createdb helios

Ensure you have Python 3.6 and pipenv installed. You can install pipenv by running pip install --user pipenv, but make sure it ends up in your PATH (e.g. add export PATH=$PATH:$HOME/.local/bin to your .bashrc).

Then, do the following:

pipenv --python python3.6
pipenv sync --dev
pipenv shell

This will create a virtualenv for you, install all the required packages (including the ones needed for development) and activate the virtual environment.

Create a local Django settings file. This will be an unversioned file that you can then customize.

cp settings/local_template.py settings/local.py

Run migrations:

python manage.py migrate

Create an institution and admin user:

python manage.py manage_users --create-institution "ZEUS"
python manage.py manage_users --create-user <username> --institution=1 --superuser

Run

python manage.py runserver 0.0.0.0:8000

Celery

By default, in development all Celery tasks run synchronously. If you want to run a Celery worker, first disable this behaviour by editing settings/local.py:

CELERY_TASK_ALWAYS_EAGER = False

You need to also install redis and make sure it's running on localhost.

Then, run:

celery worker -A zeus.celery -l INFO

Test

pytest -v

Python packages

We use pipenv to manage dependencies:

• Pipfile - contains a list of direct dependencies, with version specifiers
• Pipfile.lock - auto-generated from Pipfile, all packages, all versions pinned together with their checksums

In order to install a new package:

• pipenv install <package>
• look into Pipfile to make sure the right package was added
• test the change and commit all files