The BrightHike Stack 2017-06-28: heroku > python/django/graphene > react/redux

I started working on BrightHike, a machine learning enhanced math education plateform, on March 1st, 2017.  My last project was a Heroku/Django/Angular combination, which left me with a desire to explore a bit on the front end :-D.

Here's what the BrightHike stack currently looks like, back to front.

The back end service is Heroku, where I have a Python 3.6/Django1.11 setup that uses Heroku's built in Postgres service for persistance.  I use the Django Rest Framework for authentication, but rather than a Rest API, I use Graphene, a python implementation of GraphQL for most API querries.

Using GraphQL instead of a Rest API has been a change, but a great one.  It has allowed me to adjust the granularity and scope of a given request on the fly.  In that way it's saved a lot of code on the Python side.

There are two different front end apps, a web app and a mobile app.

I run the web app as a static app that reaches out to the back end for dynamic behavior.  The up side for this is that it makes it simple to cache the front end.  the down side is that it leaves out server side rendering.  It's possible to do some static rendering for React apps, but I haven't worried about that yet.

The web app is a standard React app, not yet even ejected from the create-react-app environment.  I'm using Redux for state management and my own twist on Ducks for organization.

The mobile app is still an unejected create-react-native-app app, which is nice for fast iterations.

The biggest changes to my setup recently have been to incorporate prettier for automatic formatting on save, and styled-components for nice in-javascript CSS handling.  I'm enjoying both a lot although I'm still working on getting my CSS architecture settled.