Shared Futures Space

A project management platform where your community collaborates and gets stuff done.

The original codebase was developed by Animorph Co-operative under community engagement lead by the consortium partners: Co-operation Ireland, Belfast Interface Project, University of Essex and Donegal Youth Service. This project was supported by the European Union’s PEACE IV Programme, managed by the Special EU Programmes Body (SEUPB).

Development

This is a Django + Wagtail + Postgres + Redis + Celery stack. We use Docker and Docker Compose to setup a development environment.

Environment

First, we need to setup the environment variables:

1. Run cp .env.example dev/.env
2. Edit dev/.env with your settings and keys

Continue with Docker section below.

Docker

This will start the codebase in dev mode, though without frontend assets. After this, continue in the section below to setup styles and JavaScript.

Linux & MINGW64 on Windows:

# build containers
USER_ID=$(id -u) GROUP_ID=$(id -g $whoami) docker compose -f dev/docker-compose.yaml up --build

macOS:

# Mac has obfuscated groups for Docker, so we use user ID
# for Dockerfile group instead of group ID
USER_ID=$(id -u) GROUP_ID=$(id -u) docker compose -f dev/docker-compose.yaml up --build

Windows (running Linux containers):

USER_ID=$(1000) GROUP_ID=$(1000) docker compose -f dev/docker-compose.yaml up --build

TypeScript

We use vite for bundling our typescript.

Make sure you have npm and nodejs installed, then install the dependencies:

npm install

To run in dev mode:

npm run dev

(or make watch will also run the same thing)

This will start a server that’s only serving static files.

Now your development instance (given you have started docker containers) is up and running! To see the app locally go to http://127.0.0.1:9000

Building assets for production

To build the files run:

npm run build

It will build the assets into sfs/vite-build.

Inside the build directory it'll put:

• manifest.json used to connect ts paths to build js files
• all the build .js files with their full path (+ a file hash)

In production vite_asset <path> will use manifest.json to lookup the path to the built .js file.

This directory is registered as a django static dir, so when collectstatic is run, it will include those resources.

The built files will be served as normal django static assets.

Commands

Enter shell:

docker compose -f dev/docker-compose.yaml exec app sh

Run Django-related administrative commands:

docker compose exec -f dev/docker-compose.yaml app django-admin startapp healerapp
# OR
docker compose exec -f dev/docker-compose.yaml app python3 manage.py startapp healerapp

Create superuser:

docker compose -f dev/docker-compose.yaml exec app python3 manage.py createsuperuser

Collect static:

docker compose -f dev/docker-compose.yaml exec app python3 manage.py collectstatic

Migrations:

docker compose -f dev/docker-compose.yaml exec app python3 manage.py makemigrations && docker compose  -f dev/docker-compose.yaml exec app python3 manage.py migrate

Running Tests:

# all tests
docker compose -f dev/docker-compose.yaml exec app pytest tests

# a specific one
docker compose -f dev/docker-compose.yaml exec app pytest tests/test_account.py
# add `-s` flag to display output
docker compose -f dev/docker-compose.yaml exec app pytest -s tests/test_account.py

Python black code formatting:

docker compose -f dev/docker-compose.yaml exec -it app make format
docker compose -f dev/docker-compose.yaml run --rm app make format

Python code linting:

docker compose -f dev/docker-compose.yaml exec -it app make lint

Development data

Fill a development database with relevant content as needed, you can load

load all data: areas, avatars, resources, organisations, users

docker compose -f dev/docker-compose.yaml exec app python3 manage.py load_devdata dev/autoupload/dev_data.json

Or only Avatar icons

docker compose -f dev/docker-compose.yaml exec app python3 manage.py load_avatars dev/autoupload/avatars.json

Or Areas

docker compose -f dev/docker-compose.yaml exec app python3 manage.py load_areas dev/autoupload/areas.json

Or Resources

docker compose -f dev/docker-compose.yaml exec app python3 manage.py load_resources dev/autoupload/resources.json

Have a look at devdata/devdata.json for some user accounts you can log in as.

Deployment

Find all the relevant information in prod/DEPLOYMENT.md

Afterwards, you might choose to add avatars for the users to choose from:

docker compose -f prod/docker-compose.prod.yaml exec app python3 manage.py load_avatars dev/autoupload/avatars.json

As well as your pre-defined areas:

docker compose -f prod/docker-compose.prod.yaml exec app python3 manage.py load_areas dev/autoupload/areas.json

And well pre-defined resources:

docker compose -f prod/docker-compose.prod.yaml exec app python3 manage.py load_resources dev/autoupload/resources.json

Resources and notes

Product terminology

This of "rivers" as work projects, "swimmers" as project members, and "springs" as areas where work can take place.

Each river is at a specific stage (envison, plan, act, reflect) at any given time. Rivers progress directionally from one stage to another.

"Resources" are knowledge resources as links. "Salmon" is our helper bot.

Directories description

• apps/: includes all Django apps
• search/: Wagtail search views
• sfs/: Django project root
• templates/: includes all Django HTML templates used across all Django apps
• tests/: Django tests using pytest

New Celery tasks

When adding celery task, restarting its container is required.

Mapping IPs onto Docker containers

We need to ensure that UID and GID from the system (host) are mapped onto the container user. The containers carry over User and Group IDs as they share one kernel. So we need to ensure that IDs of the user and group of the host match these of the container user. This behaviour is different in macOS as it automatically maps container file ownership to that of the host.

References:

• https://medium.com/@mccode/understanding-how-uid-and-gid-work-in-docker-containers-c37a01d01cf
• https://blog.dbi-services.com/how-uid-mapping-works-in-docker-containers/
• https://stackoverflow.com/a/61009413/5631104

Passing variables to Dockerfile via docker compose

Mapping the UID and GID will change depending on the environment, we don't want to change the Dockerfile each time. Does look like it's challenging/dangerous to include shell in env variables, e.g. Ref 1. Hence need to pass the user variables via CLI when building the containers.

USER_ID=$(id -u) GROUP_ID=$(id -g $whoami) docker compose up --build

Tailwind

Tailwind is include in the vite config, via PostCSS.

Note: Styles passed dynamically from views are not automatically applied to tailwind classes (which are exported as static classes at the time of save/build). So even if the classes are on the list in tailwind.config.js, but they are not used by any html element at the time of running the app you cannot refer to them.