Bundler : A gem to bundle gems
Github: http://github.com/wycats/bundler
Mailing list: http://groups.google.com/group/ruby-bundler
IRC: #carlhuda on freenode
Intro
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 bundled gems accordingly. It also lets
you run any ruby code in context of the bundled gem environment.
Installation
Bundler has no dependencies. Just clone the git repository and install the gem
with the following rake task:
rake install
You can also install the gem with
gem install bundler
Usage
Bundler requires a gem manifest file to be created. This should be a file named
Gemfile
located in the root directory of your application. After the manifest
has been created, in your shell, cd into your application's directory and run
gem bundle
. This will start the bundling process.
Manifest file
This is where you specify all of your application's dependencies. By default
this should be in a file named Gemfile
located in your application's root
directory. The following is an example of a potential Gemfile
. For more
information, please refer to Bundler::ManifestBuilder.
# Specify a dependency on rails. When the 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"
# Specify a dependency rspec, but only require that gem in the "testing"
# environment. :except is also a valid option to specify environment
# restrictions.
gem "rspec", :only => :testing
# Specify a dependency, but specify that it is already present and expanded
# at vendor/rspec. Bundler will treat rspec as though it was the rspec gem
# for the purpose of gem resolution: if another gem depends on a version
# of rspec satisfied by "1.1.6", it will be used.
#
# If a gemspec is found in the directory, it will be used to specify load
# paths and supply additional dependencies.
#
# Bundler will also recursively search for *.gemspec, and assume that
# gemspecs it finds represent gems that are rooted in the same directory
# the gemspec is found in.
gem "rspec", "1.1.6", :vendored_at => "vendor/rspec"
# Works exactly like :vendored_at, but first downloads the repo from
# git and handles stashing the files for you. As with :vendored_at,
# Bundler will automatically use *.gemspec files in the root or anywhere
# in the repository.
gem "rails", "3.0.pre", :git => "git://github.com/rails/rails.git"
# use a shallow git clone instead of a full clone
gem "rails", "2.3.4", :git => "git://github.com/rails/rails.git", :shallow => true
# Add http://gems.github.com as a source that the bundler will use
# to find gems listed in the manifest. By default,
# http://gems.rubyforge.org is already added to the list.
#
# This is an optional setting.
source "http://gems.github.com"
# Specify where the bundled gems should be stashed. This directory will
# be a gem repository where all gems are downloaded to and installed to.
#
# This is an optional setting.
# The default is: vendor/gems
bundle_path "my/bundled/gems"
# Specify where gem executables should be copied to.
#
# This is an optional setting.
# The default is: bin
bin_path "my/executables"
# Specify that rubygems should be completely disabled. This means that it
# will be impossible to require it and that available gems will be
# limited exclusively to gems that have been bundled.
#
# The default is to automatically require rubygems. There is also a
# `disable_system_gems` option that will limit available rubygems to
# the ones that have been bundled.
disable_rubygems
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".
Running Bundler
Once a manifest file has been created, the only thing that needs to be done
is to run the gem bundle
command anywhere in your application. The script
will load the manifest file, resolve all the dependencies, download all
needed gems, and install them into the specified directory.
Every time an update is made to the manifest file, run gem bundle
again to
get the changes installed. This will only check the remote sources if your
currently installed gems do not satisfy the Gemfile
. If you want to force
checking for updates on the remote sources, use the --update
option.
Remote deploys
When you run gem bundle
, the following steps occur:
- Gemfile is read in
- The gems specified in the Gemfile are resolved against the gems
already in your bundle. If the dependencies resolve, skip to step 5.
- If the dependencies in your Gemfile cannot be fully resolved
against the gems already in the bundle, the metadata for each
source is fetched.
- The gems in the Gemfile are resolved against the full list of
available gems in all sources, and the resulting gems are downloaded
- Each gem that has been downloaded but not yet expanded is expanded
into the local directory. This expansion process also installs
native gems.
As you can see, if you run gem bundle twice in a row, it will do nothing the
second time, since the gems obviously resolve against the installed gems,
and they are all expanded.
This also means that if you run gem bundle
, and .gitignore the expanded
copies, leaving only the cached .gem
files, you can run gem bundle
again
on the remote system, and it will only expand out the gems (but not
resolve or download .gem
files). This also means that native gems
will be compiled for the target platform without requiring that the
.gem
file itself be downloaded from a remote gem server.
Assuming a Rails app with Bundler's standard setup, add something like
this to your top-level .gitignore
to only keep the cache:
bin/*
vendor/gems/*
!vendor/gems/cache/
Make sure that you explicitly git add vendor/gems/cache
before you commit.
Gems with compile-time options
Some gems require you to pass compile-time options to the gem install command.
For instance, to install mysql, you might do:
gem install mysql -- --with-mysql-config=/usr/local/lib/mysql
You can pass these options to the bundler by creating a YAML file containing
the options in question:
mysql:
mysql-config: /usr/local/lib/mysql
You can then point the bundler at the file:
gem bundle --build-options build_options.yml
In general, you will want to keep the build options YAML out of version control,
and provide the appropriate options for the system in question.
Running your application
The easiest way to run your application is to start it with an executable
copied to the specified bin directory (by default, simply bin). For example,
if the application in question is a rack app, start it with bin/rackup
.
This will automatically set the gem environment correctly.
Another way to run arbitrary ruby code in context of the bundled gems is to
run it with the gem exec
command. For example:
gem exec ruby my_ruby_script.rb
You can use gem exec bash
to enter a shell that will run all binaries in
the current context.
Yet another way is to manually require the environment file first. This is
located in [bundle_path]/gems/environment.rb
. For example:
ruby -r vendor/gems/environment.rb my_ruby_script.rb
Using Bundler with Rails today
It should be possible to use Bundler with Rails today. Here are the steps
to follow.
-
In your rails app, create a Gemfile and specify the gems that your
application depends on. Make sure to specify rails as well:
gem "rails", "2.1.2"
gem "will_paginate"
# Optionally, you can disable system gems all together and only
# use bundled gems.
disable_system_gems
-
Run gem bundle
-
You can now use rails if you prepend gem exec
to every call to script/*
but that isn't fun.
-
At the top of config/preinitializer.rb
, add the following line:
require "#{RAILS_ROOT}/vendor/gems/environment"
In theory, this should be enough to get going.
To require rubygems or not
Ideally, no gem would assume the presence of rubygems at runtime. Rubygems provides
enough features so that this isn't necessary. However, there are a number of gems
that require specific rubygems features.
If the disable_rubygems
option is used, Bundler will stub out the most common
of these features, but it is possible that things will not go as intended quite
yet. So, if you are brave, try your code without rubygems at runtime.
This is different from the disable_system_gems
option, which uses the rubygems
library, but prevents system gems from being loaded; only gems that are bundled
will be available to your application. This option guarantees that dependencies
of your application will be available to a remote system.
Known Issues
- When a gem points to a git repository, the git repository will be cloned
every time Bundler does a gem dependency resolve.
Reporting bugs
Please report all bugs on the github issue tracker for the project located
at:
http://github.com/wycats/bundler/issues/