/bundler

Gemfiles are fun

Primary LanguageRubyMIT LicenseMIT

Bundler : A gem to bundle gems

Bundler is a tool that manages gem dependencies for your ruby application. It takes a gem manifest file and is able to fetch, download, and install the gems and all child dependencies specified in this manifest. It can manage any update to the gem manifest file and update the bundle's gems accordingly. It also lets you run any ruby code in context of the bundle's gem environment.

Installation

If you are upgrading from Bundler 0.8, be sure to read the upgrade notes located at the bottom of this file.

Bundler has no dependencies besides Ruby and RubyGems. You can install the latest release via RubyGems:

gem install bundler

If you want to contribute, or need a change that hasn't been released yet, just clone the git repository and install the gem with rake:

rake install

Usage

The first thing to do is create a gem manifest file named Gemfile at the root directory of your application. This can quickly be done by running bundle init in the directory that you wish the Gemfile to be created in.

Gemfile

This is where you specify all of your application's dependencies. The following is an example. For more information, refer to Bundler::Dsl.

# Add :gemcutter as a source that Bundler will use to find gems listed
# in the manifest. At least one source should be listed. URLs may also
# be used, such as http://gems.github.com.
#
source :gemcutter

# Specify a dependency on rails. When bundler downloads gems,
# it will download rails as well as all of rails' dependencies
# (such as activerecord, actionpack, etc...)
#
# At least one dependency must be specified
#
gem "rails"

# Specify a dependency on rack v.1.0.0. The version is optional.
# If present, it can be specified the same way as with rubygems'
# #gem method.
#
gem "rack", "1.0.0"

# Add a git repository as a source. Valid options include :branch, :tag,
# and :ref. Next, add any gems that you want from that repo.
#
git "git://github.com/indirect/rails3-generators.git"
gem "rails3-generators"

Groups

Applications may have dependencies that are specific to certain environments, such as testing or deployment.

You can specify groups of gems in the Gemfile using the following syntax:

gem "nokogiri", :group => :test

# or

group :test do
 gem "webrat"
end

Note that Bundler adds all the gems without an explicit group name to the :default group.

Groups are involved in a number of scenarios:

  1. When installing gems using bundle install, you can choose to leave out any group by specifying --without {group name}. This can be helpful if, for instance, you have a gem that you cannot compile in certain environments.
  2. When setting up load paths using Bundler.setup, Bundler will, by default, add the load paths for all groups. You can restrict the groups to add by doing Bundler.setup(:group, :names). If you do this, you need to specify the :default group if you want it included.
  3. When auto-requiring files using Bundler.require, Bundler will, by default, auto-require just the :default group. You can specify a list of groups to auto-require such as Bundler.require(:default, :test)

Installing gems

Once the Gemfile manifest file has been created, the next step is to install all the gems needed to satisfy the manifest's dependencies. The command to do this is bundle install.

This command will load the Gemfile, resolve all the dependencies, download all gems that are missing, and install them to the bundler's gem repository. Gems that are already installed into the system RubyGems repository will be referenced, rather than installed again. Every time an update is made to the Gemfile, run bundle install again to install any newly needed gems.

If you want to install the gems into the project's folder, like Bundler 0.8 and earlier did, you can run bundle install vendor, and the gems will be installed into the vendor subdirectory of your project.

Locking dependencies

By default, bundler will only ensure that the activated gems satisfy the Gemfile's dependencies. If you install a newer version of a gem and it satisfies the dependencies, it will be used instead of the older one.

The command bundle lock will lock the bundle to the current set of resolved gems. This ensures that, until the lock file is removed, bundle install and Bundle.setup will always activate the same gems.

When you are distributing your application, you should add the Gemfile and Gemfile.lock files to your source control, so that the set of libraries your code will run against are fixed. Simply run bundle install after checking out or deploying your code to ensure your libraries are present.

DO NOT add the .bundle directory to your source control. The files there are internal to bundler and vary between machines. If you are using git, you can exclude all machine-specific bundler files by adding a single line to your .gitignore file containing .bundle.

Running the application

Bundler must be required and setup before anything else is required. This is because it will configure all the load paths and manage gems for you. To do this, include the following at the beginning of your code.

begin
  # Try to require the preresolved locked set of gems.
  require File.expand_path('../.bundle/environment', __FILE__)
rescue LoadError
  # Fall back on doing an unlocked resolve at runtime.
  require "rubygems"
  require "bundler"
  Bundler.setup
end

# Your application's requires come here, e.g.
# require 'date' # a ruby standard library
# require 'rack' # a bundled gem

# Alternatively, you can require all the bundled libs at once
# Bundler.require

The bundle exec command provides a way to run arbitrary ruby code in context of the bundle. For example:

bundle exec ruby my_ruby_script.rb

To enter a shell that will run all gem executables (such as rake, rails, etc... ) use bundle exec bash (replacing bash for whatever your favorite shell is).

Packing the bundle's gems

When sharing or deploying an application, you may want to include everything necessary to install gem dependencies. bundle package will copy .gem files for all of the bundle's dependencies into vendor/cache. After that, bundle install will always work, since it will install the local .gem files, and will not contact any of the remote sources.

Gem resolution

One of the most important things that the bundler does is do a dependency resolution on the full list of gems that you specify, all at once. This differs from the one-at-a-time dependency resolution that Rubygems does, which can result in the following problem:

# On my system:
#   activesupport 3.0.pre
#   activesupport 2.3.4
#   activemerchant 1.4.2
#   rails 2.3.4
#
# activemerchant 1.4.2 depends on activesupport >= 2.3.2

gem "activemerchant", "1.4.2"
# results in activating activemerchant, as well as
# activesupport 3.0.pre, since it is >= 2.3.2

gem "rails", "2.3.4"
# results in:
#   can't activate activesupport (= 2.3.4, runtime)
#   for ["rails-2.3.4"], already activated
#   activesupport-3.0.pre for ["activemerchant-1.4.2"]

This is because activemerchant has a broader dependency, which results in the activation of a version of activesupport that does not satisfy a more narrow dependency.

Bundler solves this problem by evaluating all dependencies at once, so it can detect that all gems together require activesupport "2.3.4".

Upgrading from Bundler 0.8 to 0.9 and above

Upgrading to Bundler 0.9 from Bundler 0.8 requires upgrading several API calls in your Gemfile, and some workarounds if you are using Rails 2.3.

Rails 2.3

Using Bundler 0.9 with Rails 2.3 requires adding a preinitializer, and making a few changes to boot.rb. The exact changes needed can be found at http://gist.github.com/302406.

Gemfile Removals

Bundler 0.9 removes the following Bundler 0.8 Gemfile APIs:

  1. disable_system_gems: This is now the default (and only) option for bundler. Bundler uses the system gems you have specified in the Gemfile, and only the system gems you have specified (and their dependencies)
  2. disable_rubygems: This is no longer supported. We are looking into ways to get the fastest performance out of each supported scenario, and we will make speed the default where possible.
  3. clear_sources: Bundler now defaults to an empty source list. If you want to include Rubygems, you can add the source via source "http://gemcutter.org". If you use bundle init, this source will be automatically added for you in the generated Gemfile
  4. bundle_path: You can specify this setting when installing via bundle install /path/to/bundle. Bundler will remember where you installed the dependencies to on a particular machine for future installs, loads, setups, etc.
  5. bin_path: Bundler no longer generates binaries in the root of your app. You should use bundle exec to execute binaries in the current context.

Gemfile Changes

Bundler 0.9 changes the following Bundler 0.8 Gemfile APIs:

  1. Bundler 0.8 supported :only and :except as APIs for describing groups of gems. Bundler 0.9 supports a single group method, which you can use to group gems together. See the above "Group" section for more information.

    This means that gem "foo", :only => :production becomes gem "foo", :group => :production, and only :production { gem "foo" } becomes group :production { gem "foo" }

    The short version is: group your gems together logically, and use the available commands to make use of the groups you've created.

  2. :require_as becomes :require

  3. :vendored_at is fully removed; you should use :path

API Changes

  1. Bundler.require_env(:environment) becomes Bundler.require(:multiple, :groups). You must now specify the default group (the default group is the group made up of the gems not assigned to any group) explicitly. So Bundler.require_env(:test) becomes Bundler.require(:default, :test)

  2. require 'vendor/gems/environment': In unlocked mode, where using system gems, this becomes Bundler.setup(:multiple, :groups). If you don't specify any groups, this puts all groups on the load path. In locked, mode, it becomes require '.bundle/environment'

More information

Explanations of common Bundler use cases can be found in Using Bundler in Real Life. The general philosophy behind Bundler 0.9 is explained at some length in Bundler 0.9: Heading Toward 1.0.

Deploying to memory-constrained servers

When deploying to a server that is memory-constrained, like Dreamhost, you should run bundle package on your local development machine, and then check in the resulting Gemfile.lock file and vendor/cache directory. The lockfile and cached gems will mean bundler can just install the gems immediately, without contacting any gem servers or using a lot of memory to resolve the dependency tree. On the server, you only need to run bundle install after you update your deployed code.

Other questions

Any remaining questions may be asked via IRC in #carlhuda on Freenode, or via email on the Bundler mailing list.

Reporting bugs

Please report all bugs on the github issue tracker for the project, located at http://github.com/carlhuda/bundler/issues/.

The best possible scenario is a ticket with a fix for the bug and a test for the fix. If that's not possible, instructions to reproduce the issue are vitally important. If you're not sure exactly how to reproduce the issue that you are seeing, create a gist of the following information and include it in your ticket:

  • Whether you have locked or not
  • What version of bundler you are using
  • Your Gemfile
  • The command you ran to generate exception(s)
  • The exception backtrace(s)

If you are using Rails 2.3, please also include:

  • Your boot.rb file
  • Your preinitializer.rb file
  • Your environment.rb file