Heroku buildpack: Ruby

This is a Heroku buildpack for Ruby, Rack, and Rails apps. It uses Bundler for dependency management.

The Subdirectory Feature

The difference between this buildpack and heroku's standard Ruby buildpack is that you can run an app in a project subdirectory. In other words, your project doesn't need to be in the root directory; it could be in a subdirectory like web/ or rails/.

To make this work, you need to set two envoronment variables BEFORE pushing to a new heroku app: APP_SUBDIR and BUNDLE_GEMFILE.

For example, if we're deploying a Rails app that lives in a subdirectory web, you would need to set:

• APP_SUBDIR=web is the name of the subdirectory that your Rails app lives in
• BUNDLE_GEMFILE=web/Gemfile is the location of the Gemfile of your Rails app

TODO: One of these can be determined from the other; update the code to only require one environment variable.

Usage

Ruby

Example Usage:

$ ls
Gemfile Gemfile.lock

$ heroku create --stack cedar --buildpack https://github.com/makersquare/heroku-buildpack-ruby
$ heroku config:add APP_SUBDIR=web BUNDLE_GEMFILE=web/Gemfile

$ git push heroku master
...
-----> Heroku receiving push
-----> Fetching custom buildpack
-----> Ruby app detected
-----> Installing dependencies using Bundler version 1.1.rc
       Running: bundle install --without development:test --path vendor/bundle --deployment
       Fetching gem metadata from http://rubygems.org/..
       Installing rack (1.3.5)
       Using bundler (1.1.rc)
       Your bundle is complete! It was installed into ./vendor/bundle
       Cleaning up the bundler cache.
-----> Discovering process types
       Procfile declares types -> (none)
       Default types for Ruby  -> console, rake

The buildpack will detect your app as Ruby if it has a Gemfile and Gemfile.lock files in the root directory. It will then proceed to run bundle install after setting up the appropriate environment for ruby and Bundler.

Run the Tests

The tests on this buildpack are written in Rspec to allow the use of focused: true. Parallelization of testing is provided by https://github.com/grosser/parallel_tests this lib spins up an arbitrary number of processes and running a different test file in each process, it does not parallelize tests within a test file. To run the tests: clone the repo, then bundle install then clone the test fixtures by running:

$ hatchet install

Now run the tests:

$ bundle exec parallel_rspec -n 6 spec/

If you don't want to run them in parallel you can still:

$ bundle exec rake spec

Now go take a nap or something for a really long time.

Bundler

For non-windows Gemfile.lock files, the --deployment flag will be used. In the case of windows, the Gemfile.lock will be deleted and Bundler will do a full resolve so native gems are handled properly. The vendor/bundle directory is cached between builds to allow for faster bundle install times. bundle clean is used to ensure no stale gems are stored between builds.

Rails 2

Example Usage:

$ ls
app  config  db  doc  Gemfile  Gemfile.lock  lib  log  public  Rakefile  README  script  test  tmp  vendor

$ ls config/environment.rb
config/environment.rb

$ heroku create --stack cedar --buildpack https://github.com/makersquare/heroku-buildpack-ruby.git

$ git push heroku master
...
-----> Heroku receiving push
-----> Ruby/Rails app detected
-----> Installing dependencies using Bundler version 1.1.rc
...
-----> Writing config/database.yml to read from DATABASE_URL
-----> Rails plugin injection
       Injecting rails_log_stdout
-----> Discovering process types
       Procfile declares types      -> (none)
       Default types for Ruby/Rails -> console, rake, web, worker

The buildpack will detect your app as a Rails 2 app if it has a environment.rb file in the config directory.

Rails Log STDOUT

A rails_log_stdout is installed by default so Rails' logger will log to STDOUT and picked up by Heroku's logplex.

Auto Injecting Plugins

Any vendored plugin can be stopped from being installed by creating the directory it's installed to in the slug. For instance, to prevent rails_log_stdout plugin from being injected, add vendor/plugins/rails_log_stdout/.gitkeep to your git repo.

Rails 3

Example Usage:

$ ls
app  config  config.ru  db  doc  Gemfile  Gemfile.lock  lib  log  Procfile  public  Rakefile  README  script  tmp  vendor

$ ls config/application.rb
config/application.rb

$ heroku create --stack cedar --buildpack https://github.com/makersquare/heroku-buildpack-ruby.git

$ git push heroku master
-----> Heroku receiving push
-----> Ruby/Rails app detected
-----> Installing dependencies using Bundler version 1.1.rc
       Running: bundle install --without development:test --path vendor/bundle --deployment
       ...
-----> Writing config/database.yml to read from DATABASE_URL
-----> Preparing app for Rails asset pipeline
       Running: rake assets:precompile
-----> Rails plugin injection
       Injecting rails_log_stdout
       Injecting rails3_serve_static_assets
-----> Discovering process types
       Procfile declares types      -> web
       Default types for Ruby/Rails -> console, rake, worker

The buildpack will detect your apps as a Rails 3 app if it has an application.rb file in the config directory.

Assets

To enable static assets being served on the dyno, rails3_serve_static_assets is installed by default. If the execjs gem is detected then node.js will be vendored. The assets:precompile rake task will get run if no public/manifest.yml is detected. See this article on how rails 3.1 works on cedar.

Hacking

To use this buildpack, fork it on Github. Push up changes to your fork, then create a test app with --buildpack <your-github-url> and push to it.

To change the vendored binaries for Bundler, Node.js, and rails plugins, use the rake tasks provided by the Rakefile. You'll need an S3-enabled AWS account and a bucket to store your binaries in as well as the vulcan gem to build the binaries on heroku.

For example, you can change the vendored version of Bundler to 1.1.rc.

First you'll need to build a Heroku-compatible version of Node.js:

$ export AWS_ID=xxx AWS_SECRET=yyy S3_BUCKET=zzz
$ s3 create $S3_BUCKET
$ rake gem:install[bundler,1.1.rc]

Open lib/language_pack/ruby.rb in your editor, and change the following line:

BUNDLER_VERSION = "1.1.rc"

Open lib/language_pack/base.rb in your editor, and change the following line:

VENDOR_URL = "https://s3.amazonaws.com/zzz"

Commit and push the changes to your buildpack to your Github fork, then push your sample app to Heroku to test. You should see:

-----> Installing dependencies using Bundler version 1.1.rc

NOTE: You'll need to vendor the plugins, node, Bundler, and libyaml by running the rake tasks for the buildpack to work properly.

Flow

Here's the basic flow of how the buildpack works:

Ruby (Gemfile and Gemfile.lock is detected)

• runs Bundler
• installs binaries
  • installs node if the gem execjs is detected
• runs rake assets:precompile if the rake task is detected

Rack (config.ru is detected)

• everything from Ruby
• sets RACK_ENV=production

Rails 2 (config/environment.rb is detected)

• everything from Rack
• sets RAILS_ENV=production
• install rails 2 plugins
  • rails_log_stdout

Rails 3 (config/application.rb is detected)

• everything from Rails 2
• install rails 3 plugins
  • rails3_server_static_assets