Represent Minnesota

Minnesota based instance of Represent Boundaries.

Installing

1. Follow this documentation to setup Python and PostGIS. As of July 16, 2019, from this URL you should start by following steps 1-7 under Install Dependencies, and then step 1 under Install Represent Boundaries.
2. Clone this git repository.
3. Run pipenv install. This completes steps 2-6 of the Install Represent Boundaries section of the docs above.
4. Copy .env.example to .env and fill in the values from your local setup (or Heroku if you are deploying). This veers away from the docs above for the sake of hosting on Heroku more easily. SECRET_KEY cannot be blank.
5. Run python manage.py migrate to initially populate the database.

Running

1. pipenv shell
2. python manage.py loadshapefiles (use --reload to reload the shapefiles)
3. python manage.py runserver

Heroku setup

1. Run heroku pg:psql to log in to your Heroku Postgres database.
2. Run CREATE EXTENSION postgis; to install the PostGIS extension to Postgres.
3. Run exit; to get out of Postgres.
4. Run heroku config:set BUILD_WITH_GEO_LIBRARIES=1 to instruct Heroku to build with geo libraries (this potentially takes a long time). Otherwise you'll run into issues when you try the next step.
5. Run python manage.py loadshapefiles to populate the data Represent Boundaries will use (for our quantity of data, this takes a long time).

Installing new boundaries

1. Find the dataset. If it's a redistricting of an existing boundary, you can often find it at the same place, with changing the URL for the year.
2. Download the shapefiles and put them into a subfolder under data/shapefiles. Update data/shapefiles/definitions.py to match the new file path and whatever information you have from the source.
3. With updated files, things like name_func, id_func, and slug_func will sometimes need new parameters. Sometimes this information is available in metadata for the new set.
4. After pushing to Heroku with the new shapefiles, follow the steps below for running on Heroku.

When redistricting happens

It can be hard to find updated data sources. The state may release a Redistricting boundary set, like this one from 2020, which "includes MN Secretary of State election data from 2012-2020 that was disaggregated to census block level and aggregated to voting districts, minor civil divisions and counties." In this case, it's possible to separate census blocks, counties, minor civil divisions, and voting districts and run them within separate folders.

Running on Heroku

1. Deploy production code to Heroku by running git push heroku master. The production version deploys to https://represent-minnesota.herokuapp.com/.
2. Deploy staging code to Heroku by pushing to the staging branch of this repository. This verison deploys at https://represent-minnesota-staging.herokuapp.com/.
3. To update the shapefiles in the Heroku database, run heroku run python manage.py loadshapefiles --reload (use --reload to reload the shapefiles). Currently this should only happen with the full dataset in production and not staging, since the set is too big for the free database plan.

Working with Django admin

1. It can be useful to have a login for the Django admin. To create one locally, run python manage.py createsuperuser. To do it on Heroku, run heroku run python manage.py createsuperuser --app represent-minnesota.
2. Boundaries and boundary sets can both be modified via the Django admin.