Spring is a Rails application preloader. It's trying to solve the same problem as spork, zeus and commands.
I made it because we are having a discussion on the rails core team about shipping something to solve this problem with rails. So this is my proposal, as working code.
(At least I hope it's working code, but this is alpha software at the moment. Please do try it and let me know if you hit problems.)
Spring is most similar to Zeus, but it's implemented in pure Ruby, and is more tightly integrated with Rails (it makes use of Rails' built-in code reloader).
Spring tries to be totally automatic.
It boots up in the background the first time you run a
command. Then it speeds up subsequent commands. If it detects that your
pre-loaded environment has changed (maybe config/application.rb
has
been edited) then it will reload your environment in the background,
ready for the next command. When you close your terminal session, Spring
will automatically shut down. There's no "server" to manually start and
stop.
Spring operates via a command line interface. Other solutions (e.g. commands) take the approach of using a special console to run commands from. This means we will have to re-implement shell features such as history, completion, etc. Whilst it's not impossible to re-implement those features, it's unnecessary work and our re-implementation won't be as feature complete as a real shell. Using a real shell also prevents the user having to constantly jump between a terminal with a real shell and a terminal running the rails "commands console".
Ruby versions supported:
- MRI 1.9.3
- MRI 2.0.0
Rails versions supported:
- 3.2
Spring makes extensive use of Process#fork
, so won't be able to run on
any platform which doesn't support that (Windows, JRuby).
Install the spring
gem. You can add it to your Gemfile if you like but
it's optional. You now have a spring
command. Don't use it with
bundle exec
or it will be extremely slow.
For this walkthrough, I'm using the test app in the Spring repository:
cd /path/to/spring/test/apps/rails-3-2
We can run a test:
$ time spring testunit test/functional/posts_controller_test.rb
Run options:
# Running tests:
.......
Finished tests in 0.127245s, 55.0121 tests/s, 78.5887 assertions/s.
7 tests, 10 assertions, 0 failures, 0 errors, 0 skips
real 0m2.165s
user 0m0.281s
sys 0m0.066s
That booted our app in the background:
$ spring status
Spring is running:
26150 spring server | rails-3-2 | started 3 secs ago
26155 spring app | rails-3-2 | started 3 secs ago | test mode
We can see two processes, one is the Spring server, the other is the application running in the test environment. When we close the terminal, the processes will be killed automatically.
Running the test is faster next time:
$ time spring testunit test/functional/posts_controller_test.rb
Run options:
# Running tests:
.......
Finished tests in 0.176896s, 39.5714 tests/s, 56.5305 assertions/s.
7 tests, 10 assertions, 0 failures, 0 errors, 0 skips
real 0m0.610s
user 0m0.276s
sys 0m0.059s
Running spring testunit
, spring rake
, spring rails
, etc gets a bit
tedious. It also suffers from a performance issue in Rubygems (which I
am actively working on)
which means the spring
command takes a while to start up. The more
gems you have, the longer it takes.
Spring binstubs solve both of these problems. If you will be running the
testunit
command regularly, run:
$ spring binstub testunit
This generates a bin/spring
and a bin/testunit
, which allows you to run
spring
and spring testunit
in a way that doesn't trigger the Rubygems
performance bug:
$ time bin/testunit test/functional/posts_controller_test.rb
Run options:
# Running tests:
.......
Finished tests in 0.166585s, 42.0207 tests/s, 60.0296 assertions/s.
7 tests, 10 assertions, 0 failures, 0 errors, 0 skips
real 0m0.407s
user 0m0.077s
sys 0m0.059s
You can add "./bin" to your PATH
when in your application's directory
with direnv, but you should
recognise and understand the security implications of using that.
If we edit any of the application files, or test files, the change will be picked up on the next run, without the background process having to be restarted.
If we edit any of the preloaded files, the application needs to restart
automatically. Let's "edit" config/application.rb
:
$ touch config/application.rb
$ spring status
Spring is running:
26150 spring server | rails-3-2 | started 36 secs ago
26556 spring app | rails-3-2 | started 1 sec ago | test mode
The application process detected the change and exited. The server process then detected that the application process exited, so it started a new application. All of this happened automatically. Next time we run a command we'll be running against a fresh application. We can see that the start time and PID of the app process has changed.
If we run a command that uses a different environment, then it gets
booted up. For example, the rake
command uses the development
environment by default:
$ spring binstub rake
$ bin/rake routes
posts GET /posts(.:format) posts#index
POST /posts(.:format) posts#create
new_post GET /posts/new(.:format) posts#new
edit_post GET /posts/:id/edit(.:format) posts#edit
post GET /posts/:id(.:format) posts#show
PUT /posts/:id(.:format) posts#update
DELETE /posts/:id(.:format) posts#destroy
We now have 3 processes: the server, and application in test mode and the application in development mode.
$ bin/spring status
Spring is running:
26150 spring server | rails-3-2 | started 1 min ago
26556 spring app | rails-3-2 | started 42 secs ago | test mode
26707 spring app | rails-3-2 | started 2 secs ago | development mode
To stop the background processes:
$ bin/spring stop
Spring stopped.
The following commands are shipped by default.
Custom commands can be specified in config/spring.rb
. See
lib/spring/commands.rb
for examples.
A bunch of different test frameworks are supported at the moment in order to make it easy for people to try spring. However in the future the code to use a specific test framework should not be contained in the spring repository.
Runs a test (e.g. Test::Unit, MiniTest::Unit, etc.) Preloads the test_helper
file.
This command can also recursively run a directory of tests. For example,
spring testunit test/functional
will run test/functional/**/*_test.rb
.
Runs an rspec spec, exactly the same as the rspec
executable. Preloads
the spec_helper
file.
Runs a cucumber feature.
Runs a rake task. Rake tasks run in the development
environment by
default. You can change this on the fly by using the RAILS_ENV
environment variable. The environment is also configurable with the
Spring::Commands::Rake.environment_matchers
hash. This has sensible
defaults, but if you need to match a specific task to a specific
environment, you'd do it like this:
Spring::Commands::Rake.environment_matchers["perf_test"] = "test"
Spring::Commands::Rake.environment_matchers[/^perf/] = "test"
These execute the rails command you already know and love. If you run
a different sub command (e.g. rails server
) then spring will automatically
pass it through to the underlying rails
executable (without the
speed-up).
Spring will read config/spring.rb
for custom settings, described below.
Spring must know how to find your Rails application. If you have a normal app everything works out of the box. If you are working on a project with a special setup (an engine for example), you must tell Spring where your app is located:
Spring.application_root = './test/dummy'
Every Spring command has the ability to preload a set of files. The
test
command for example preloads test_helper
(it also adds the
test/
directory to your load path). If the
defaults don't work for your application you can configure the
preloads for every command:
# if your test helper is called "helper"
Spring::Commands::TestUnit.preloads = %w(helper)
# if you don't want to preload spec_helper.rb
Spring::Commands::RSpec.preloads = []
# if you want to preload additional files for the console
Spring::Commands::RailsConsole.preloads << 'extensions/console_helper'
You might want to run code after Spring forked off the process but
before the actual command is run. You might want to use an
after_fork
callback if you have to connect to an external service,
do some general cleanup or set up dynamic configuration.
Spring.after_fork do
# run arbitrary code
end
If you want to register multiple callbacks you can simply call
Spring.after_fork
multiple times with different blocks.
As mentioned above, Spring will automatically detect file changes to any file loaded when the server
boots. If you would like to watch additional files or directories, use watch
:
Spring.watch "spec/factories"
Spring needs a tmp directory. This will default to Rails.root.join('tmp', 'spring')
.
You can set your own configuration directory by setting the SPRING_TMP_PATH
environment variable.
By default Spring will check the filesystem for changes once every 0.2 seconds. This
method requires zero configuration, but if you find that it's using too
much CPU, then you can turn on event-based file system listening by
adding the following to to your Gemfile
:
group :development, :test do
gem 'listen'
gem 'rb-inotify', :require => false # linux
gem 'rb-fsevent', :require => false # mac os x
gem 'rb-kqueue', :require => false # bsd
end
Note that this make the initial application startup slightly slower.