Chris Langager

Environments

It's very common to have different environments that your code will run in. The three most common that I've seen are:

development - this is a messy place where can break things and learn

staging - a replication of production, used to test before rolling out changes to production

production - what your end users are using

As you write code to fix bugs and make new features, you will update each of these environments with your changes. Though your code will not change from environment to environment, the configuration of that environment will. Typically, a major part of this config is handled by environment variables. They might look something like this:

ENVIRONMENT=staging
DATABASE_CONNECTION=postgresql://user:[email protected]/postgres
PORT=8080

ENVIRONMENT=production
DATABASE_CONNECTION=postgresql://user:[email protected]/postgres
PORT=8080

These envrionment variables can now be used to change the configuration/behavior of your code. Here's what that might look like for an API backed by a postgres database:

db, err := sql.Open("postgres", os.Getenv("POSTGRES_URL"))
if err != nil {
    log.Fatalf("unable to connect to db with url %s: %v", os.Getenv("POSTGRES_URL"), err)
}

mux := http.NewServeMux()
mux.HandleFunc("GET /users", listUsers(db))
mux.HandleFunc("GET /users/{id}", getUser(db))

port := os.Getenv("PORT")
log.Printf("listening on :%s...\n", port)
err := http.ListenAndServe(fmt.Sprintf(":%s", port), mux)
if err != nil {
    log.Fatalf("unable to start server: %v", err)
}

Now, let's say you are tasked with adding a new endpoint for creating users, but we aren't ready to have that released in production. You might be tempted to do something like this:

var postgresUrl string
if os.Getenv("ENVIRONMENT") != "production" {
    mux.HandleFunc("POST /users", createUser(db))
}

Instead, do this:

var postgresUrl string
if os.Getenv("CREATE_USERS_ENABLED") == "true" {
    mux.HandleFunc("POST /users", createUser(db))
}

Software