Rails on Heroku: git add environment.rb or not? - ruby-on-rails

Question

I usually git-ignore config/environment.rb file. I add ENV['RAILS_ENV'] setting into that file so it should be different on different machine.
However, Heroku requires that file in git. Without it, the application crashes.
How do you handle it?
Thanks.
Sam

Answer 1

You shouldn't ignore config/environment.rb. As Zeke already answered, you can set environment variables in a live Heroku app.
To see a list of environment variables available to your heroku app run heroku config. To set the value of an specific variable run heroku config:add RAILS_ENV=staging --app yourapp.
But what about your development app running in your own machine? There are some options, one is exporting an environment variable in your shell:
$ export RAILS_ENV=staging
$ echo $RAILS_ENV
staging
But I prefer another approach. This is what I have in my config/environment.rb:
# Load the rails application
require File.expand_path('../application', __FILE__)
# Load environment vars from local file
env_vars = File.join(Rails.root.to_s, 'config', 'env_vars.rb')
load(env_vars) if File.exists?(env_vars)
# Initialize the rails application
MyApp::Application.initialize!
If there is a config/env_vars.rb file, it will be loaded. This is the contents of a sample env_vars.rb file:
ENV["REDISTOGO_URL"] = 'redis://localhost:6379'
ENV["AWS_ACCESS_KEY"] = '283983483'
ENV["AWS_SECRET"] = '743843934'
ENV["S3_BUCKET"] = 'myapp-development'
This way you could have different sets of environment variables for your development, staging and production apps.
You will want to git-ignore this file.
While I'm running my development app I want to send files to a myapp-development S3 bucket. But I want my production app (in Heroku) to send files to a myapp-production S3 bucket.
$ heroku config:add S3_BUCKET='myapp-production' --app myapp
And a different bucket for my staging app (also running in Heroku):
$ heroku config:add S3_BUCKET='myapp-staging' --app myapp-staging
Since you'll git-ignore config/env_vars.rb, it will not be present in your apps running in Heroku, and so they'll get their environment variables from the above heroku config:add configurations.
Since I work with other people, I often include an example env_vars.rb.example file in the project, which is not ignored by git:
File config/env_vars.rb.example:
# Copy this file to config/env_vars.rb
# MAKE SURE THAT YOUR .gitignore INCLUDES IT!
#
# This file is used to set environment variables that would be present on
# Heroku, and are needed for our development and test instances to run.
ENV["REDISTOGO_URL"] = 'redis://localhost:6379'
ENV["AWS_ACCESS_KEY"] = 'xxx'
ENV["AWS_SECRET"] = 'xxx'
ENV["S3_BUCKET"] = 'some-bucket'
if Rails.env.test?
HOST = "example.com"
else
HOST = "development.intranet"
end
Notice that an advantage of loading this env_vars.rb file is that you can do different things programmatically, for example setting the HOST constant to a different value if you are running your tests, like above.

Answer 2

Heroku has pretty great documentation. To answer your question directly, they provide and interface to configure environment variables. They even provide a mechanism that makes deploying to two+ Heroku-hosted apps with different environments quite easy.