At the time of building this, I am a Junior Scala Developer. About half a year ago I started a new job where the company programmes in Scala with a functional tech stack.

The idea of this app is simple and follow CRUD format:

1. POST a book
2. GET a book
3. PUT (update) a book
4. DELETE a book

The core purpose of app one is for me to understand how to deploy a very simple app.

For this app I deployed using Heroku, I chose to use this platform as I thought it would be an easy place to start. In the future, I will try to deploy to AWS in dockerised containers.

• Build a simple backend Scala app
• connect a database
• write a few tests
• deploy to Herko
• Add CI/CD

• Scala
• Cats
• Cats Effect
• Doobie
• Docker-compose
• Circe
• Heroku

I wasn't too familar on how to properly use environment variables and application.conf files.

My original plan was to have an application.conf file and have a meta directory. The meta directory would have two files: dev.env and prd.env. I would have something which ran source meta/prd.env when in production (heroku).

Unfortunately this cannot be achieved as I could not store variables statically in that location/file as Heroku updates its variables every so often, meaning the variables within prd.env would become out of date.

Addtionally my original plan would be against the https://www.12factor.net/ coding principles.

 def prodConfig(): Config = {
    val dbUri = new URI(System.getenv("DATABASE_URL"))
    val username = dbUri.getUserInfo.split(":")(0)
    val password = dbUri.getUserInfo.split(":")(1)
    val dbUrl = "jdbc:postgresql://" + dbUri.getHost + dbUri.getPath

    Config(ServerConfig(5432, dbUri.getHost), DbConfig(dbUrl, username, password, 10))
  }

Instead of using LoadConfig I used the above function to capture and chop the DATABASE_URL which is injected by Heroku, I then pass that to the various parts of config that need it.