**DO NOT READ THIS FILE ON GITHUB, GUIDES ARE PUBLISHED ON http://guides.rubyonrails.org.** Ruby on Rails 4.2 Release Notes =============================== Highlights in Rails 4.2: * Active Job * Asynchronous mails * Adequate Record * Web Console * Foreign key support These release notes cover only the major changes. To learn about other features, bug fixes, and changes, please refer to the changelogs or check out the [list of commits](https://github.com/rails/rails/commits/4-2-stable) in the main Rails repository on GitHub. -------------------------------------------------------------------------------- Upgrading to Rails 4.2 ---------------------- If you're upgrading an existing application, it's a great idea to have good test coverage before going in. You should also first upgrade to Rails 4.1 in case you haven't and make sure your application still runs as expected before attempting to upgrade to Rails 4.2. A list of things to watch out for when upgrading is available in the guide [Upgrading Ruby on Rails](upgrading_ruby_on_rails.html#upgrading-from-rails-4-1-to-rails-4-2). Major Features -------------- ### Active Job Active Job is a new framework in Rails 4.2. It is a common interface on top of queuing systems like [Resque](https://github.com/resque/resque), [Delayed Job](https://github.com/collectiveidea/delayed_job), [Sidekiq](https://github.com/mperham/sidekiq), and more. Jobs written with the Active Job API run on any of the supported queues thanks to their respective adapters. Active Job comes pre-configured with an inline runner that executes jobs right away. Jobs often need to take Active Record objects as arguments. Active Job passes object references as URIs (uniform resource identifiers) instead of marshaling the object itself. The new [Global ID](https://github.com/rails/globalid) library builds URIs and looks up the objects they reference. Passing Active Record objects as job arguments just works by using Global ID internally. For example, if `trashable` is an Active Record object, then this job runs just fine with no serialization involved: ```ruby class TrashableCleanupJob < ActiveJob::Base def perform(trashable, depth) trashable.cleanup(depth) end end ``` See the [Active Job Basics](active_job_basics.html) guide for more information. ### Asynchronous Mails Building on top of Active Job, Action Mailer now comes with a `deliver_later` method that sends emails via the queue, so it doesn't block the controller or model if the queue is asynchronous (the default inline queue blocks). Sending emails right away is still possible with `deliver_now`. ### Adequate Record Adequate Record is a set of performance improvements in Active Record that makes common `find` and `find_by` calls and some association queries up to 2x faster. It works by caching common SQL queries as prepared statements and reusing them on similar calls, skipping most of the query-generation work on subsequent calls. For more details, please refer to [Aaron Patterson's blog post](http://tenderlovemaking.com/2014/02/19/adequaterecord-pro-like-activerecord.html). Active Record will automatically take advantage of this feature on supported operations without any user involvement or code changes. Here are some examples of supported operations: ```ruby Post.find(1) # First call generates and cache the prepared statement Post.find(2) # Subsequent calls reuse the cached prepared statement Post.find_by_title('first post') Post.find_by_title('second post') Post.find_by(title: 'first post') Post.find_by(title: 'second post') post.comments post.comments(true) ``` It's important to highlight that, as the examples above suggest, the prepared statements do not cache the values passed in the method calls; rather, they have placeholders for them. Caching is not used in the following scenarios: - The model has a default scope - The model uses single table inheritance - `find` with a list of ids, e.g.: ```ruby # not cached Post.find(1, 2, 3) Post.find([1,2]) ``` - `find_by` with SQL fragments: ```ruby Post.find_by('published_at < ?', 2.weeks.ago) ``` ### Web Console New applications generated with Rails 4.2 now come with the [Web Console](https://github.com/rails/web-console) gem by default. Web Console adds an interactive Ruby console on every error page and provides a `console` view and controller helpers. The interactive console on error pages lets you execute code in the context of the place where the exception originated. The `console` helper, if called anywhere in a view or controller, launches an interactive console with the final context, once rendering has completed. ### Foreign Key Support The migration DSL now supports adding and removing foreign keys. They are dumped to `schema.rb` as well. At this time, only the `mysql`, `mysql2` and `postgresql` adapters support foreign keys. ```ruby # add a foreign key to `articles.author_id` referencing `authors.id` add_foreign_key :articles, :authors # add a foreign key to `articles.author_id` referencing `users.lng_id` add_foreign_key :articles, :users, column: :author_id, primary_key: "lng_id" # remove the foreign key on `accounts.branch_id` remove_foreign_key :accounts, :branches # remove the foreign key on `accounts.owner_id` remove_foreign_key :accounts, column: :owner_id ``` See the API documentation on [add_foreign_key](http://api.rubyonrails.org/v4.2.0/classes/ActiveRecord/ConnectionAdapters/SchemaStatements.html#method-i-add_foreign_key) and [remove_foreign_key](http://api.rubyonrails.org/v4.2.0/classes/ActiveRecord/ConnectionAdapters/SchemaStatements.html#method-i-remove_foreign_key) for a full description. Incompatibilities ----------------- Previously deprecated functionality has been removed. Please refer to the individual components for new deprecations in this release. The following changes may require immediate action upon upgrade. ### `render` with a String Argument Previously, calling `render "foo/bar"` in a controller action was equivalent to `render file: "foo/bar"`. In Rails 4.2, this has been changed to mean `render template: "foo/bar"` instead. If you need to render a file, please change your code to use the explicit form (`render file: "foo/bar"`) instead. ### `respond_with` / Class-Level `respond_to` `respond_with` and the corresponding class-level `respond_to` have been moved to the [responders](https://github.com/plataformatec/responders) gem. Add `gem 'responders', '~> 2.0'` to your `Gemfile` to use it: ```ruby # app/controllers/users_controller.rb class UsersController < ApplicationController respond_to :html, :json def show @user = User.find(params[:id]) respond_with @user end end ``` Instance-level `respond_to` is unaffected: ```ruby # app/controllers/users_controller.rb class UsersController < ApplicationController def show @user = User.find(params[:id]) respond_to do |format| format.html format.json { render json: @user } end end end ``` ### Default Host for `rails server` Due to a [change in Rack](https://github.com/rack/rack/commit/28b014484a8ac0bbb388e7eaeeef159598ec64fc), `rails server` now listens on `localhost` instead of `0.0.0.0` by default. This should have minimal impact on the standard development workflow as both http://127.0.0.1:3000 and http://localhost:3000 will continue to work as before on your own machine. However, with this change you will no longer be able to access the Rails server from a different machine, for example if your development environment is in a virtual machine and you would like to access it from the host machine. In such cases, please start the server with `rails server -b 0.0.0.0` to restore the old behavior. If you do this, be sure to configure your firewall properly such that only trusted machines on your network can access your development server. ### Changed status option symbols for `render` Due to a [change in Rack](https://github.com/rack/rack/commit/be28c6a2ac152fe4adfbef71f3db9f4200df89e8), the symbols that the `render` method accepts for the `:status` option have changed: - 306: `:reserved` has been removed. - 413: `:request_entity_too_large` has been renamed to `:payload_too_large`. - 414: `:request_uri_too_long` has been renamed to `:uri_too_long`. - 416: `:requested_range_not_satisfiable` has been renamed to `:range_not_satisfiable`. Keep in mind that if calling `render` with an unknown symbol, the response status will default to 500. ### HTML Sanitizer The HTML sanitizer has been replaced with a new, more robust, implementation built upon [Loofah](https://github.com/flavorjones/loofah) and [Nokogiri](https://github.com/sparklemotion/nokogiri). The new sanitizer is more secure and its sanitization is more powerful and flexible. Due to the new algorithm, the sanitized output may be different for certain pathological inputs. If you have a particular need for the exact output of the old sanitizer, you can add the [rails-deprecated_sanitizer](https://github.com/kaspth/rails-deprecated_sanitizer) gem to the `Gemfile`, to have the old behavior. The gem does not issue deprecation warnings because it is opt-in. `rails-deprecated_sanitizer` will be supported for Rails 4.2 only; it will not be maintained for Rails 5.0. See [this blog post](http://blog.plataformatec.com.br/2014/07/the-new-html-sanitizer-in-rails-4-2/) for more details on the changes in the new sanitizer. ### `assert_select` `assert_select` is now based on [Nokogiri](https://github.com/sparklemotion/nokogiri). As a result, some previously-valid selectors are now unsupported. If your application is using any of these spellings, you will need to update them: * Values in attribute selectors may need to be quoted if they contain non-alphanumeric characters. ```ruby # before a[href=/] a[href$=/] # now a[href="/"] a[href$="/"] ``` * DOMs built from HTML source containing invalid HTML with improperly nested elements may differ. For example: ```ruby # content:
AT&T
# before: assert_select('p', 'AT&T') # => true assert_select('p', 'AT&T') # => false # now: assert_select('p', 'AT&T') # => true assert_select('p', 'AT&T') # => false ``` Furthermore substitutions have changed syntax. Now you have to use a `:match` CSS-like selector: ```ruby assert_select ":match('id', ?)", 'comment_1' ``` Additionally Regexp substitutions look different when the assertion fails. Notice how `/hello/` here: ```ruby assert_select(":match('id', ?)", /hello/) ``` becomes `"(?-mix:hello)"`: ``` Expected at least 1 element matching "div:match('id', "(?-mix:hello)")", found 0.. Expected 0 to be >= 1. ``` See the [Rails Dom Testing](https://github.com/rails/rails-dom-testing/tree/8798b9349fb9540ad8cb9a0ce6cb88d1384a210b) documentation for more on `assert_select`. Railties -------- Please refer to the [Changelog][railties] for detailed changes. ### Removals * The `--skip-action-view` option has been removed from the app generator. ([Pull Request](https://github.com/rails/rails/pull/17042)) * The `rails application` command has been removed without replacement. ([Pull Request](https://github.com/rails/rails/pull/11616)) ### Deprecations * Deprecated missing `config.log_level` for production environments. ([Pull Request](https://github.com/rails/rails/pull/16622)) * Deprecated `rake test:all` in favor of `rake test` as it now run all tests in the `test` folder. ([Pull Request](https://github.com/rails/rails/pull/17348)) * Deprecated `rake test:all:db` in favor of `rake test:db`. ([Pull Request](https://github.com/rails/rails/pull/17348)) * Deprecated `Rails::Rack::LogTailer` without replacement. ([Commit](https://github.com/rails/rails/commit/84a13e019e93efaa8994b3f8303d635a7702dbce)) ### Notable changes * Introduced `web-console` in the default application Gemfile. ([Pull Request](https://github.com/rails/rails/pull/11667)) * Added a `required` option to the model generator for associations. ([Pull Request](https://github.com/rails/rails/pull/16062)) * Introduced the `x` namespace for defining custom configuration options: ```ruby # config/environments/production.rb config.x.payment_processing.schedule = :daily config.x.payment_processing.retries = 3 config.x.super_debugger = true ``` These options are then available through the configuration object: ```ruby Rails.configuration.x.payment_processing.schedule # => :daily Rails.configuration.x.payment_processing.retries # => 3 Rails.configuration.x.super_debugger # => true ``` ([Commit](https://github.com/rails/rails/commit/611849772dd66c2e4d005dcfe153f7ce79a8a7db)) * Introduced `Rails::Application.config_for` to load a configuration for the current environment. ```ruby # config/exception_notification.yml: production: url: http://127.0.0.1:8080 namespace: my_app_production development: url: http://localhost:3001 namespace: my_app_development # config/environments/production.rb Rails.application.configure do config.middleware.use ExceptionNotifier, config_for(:exception_notification) end ``` ([Pull Request](https://github.com/rails/rails/pull/16129)) * Introduced a `--skip-turbolinks` option in the app generator to not generate turbolinks integration. ([Commit](https://github.com/rails/rails/commit/bf17c8a531bc8059d50ad731398002a3e7162a7d)) * Introduced a `bin/setup` script as a convention for automated setup code when bootstrapping an application. ([Pull Request](https://github.com/rails/rails/pull/15189)) * Changed the default value for `config.assets.digest` to `true` in development. ([Pull Request](https://github.com/rails/rails/pull/15155)) * Introduced an API to register new extensions for `rake notes`. ([Pull Request](https://github.com/rails/rails/pull/14379)) * Introduced an `after_bundle` callback for use in Rails templates. ([Pull Request](https://github.com/rails/rails/pull/16359)) * Introduced `Rails.gem_version` as a convenience method to return `Gem::Version.new(Rails.version)`. ([Pull Request](https://github.com/rails/rails/pull/14101)) Action Pack ----------- Please refer to the [Changelog][action-pack] for detailed changes. ### Removals * `respond_with` and the class-level `respond_to` have been removed from Rails and moved to the `responders` gem (version 2.0). Add `gem 'responders', '~> 2.0'` to your `Gemfile` to continue using these features. ([Pull Request](https://github.com/rails/rails/pull/16526), [More Details](http://guides.rubyonrails.org/upgrading_ruby_on_rails.html#responders)) * Removed deprecated `AbstractController::Helpers::ClassMethods::MissingHelperError` in favor of `AbstractController::Helpers::MissingHelperError`. ([Commit](https://github.com/rails/rails/commit/a1ddde15ae0d612ff2973de9cf768ed701b594e8)) ### Deprecations * Deprecated the `only_path` option on `*_path` helpers. ([Commit](https://github.com/rails/rails/commit/aa1fadd48fb40dd9396a383696134a259aa59db9)) * Deprecated `assert_tag`, `assert_no_tag`, `find_tag` and `find_all_tag` in favor of `assert_select`. ([Commit](https://github.com/rails/rails-dom-testing/commit/b12850bc5ff23ba4b599bf2770874dd4f11bf750)) * Deprecated support for setting the `:to` option of a router to a symbol or a string that does not contain a "#" character: ```ruby get '/posts', to: MyRackApp => (No change necessary) get '/posts', to: 'post#index' => (No change necessary) get '/posts', to: 'posts' => get '/posts', controller: :posts get '/posts', to: :index => get '/posts', action: :index ``` ([Commit](https://github.com/rails/rails/commit/cc26b6b7bccf0eea2e2c1a9ebdcc9d30ca7390d9)) * Deprecated support for string keys in URL helpers: ```ruby # bad root_path('controller' => 'posts', 'action' => 'index') # good root_path(controller: 'posts', action: 'index') ``` ([Pull Request](https://github.com/rails/rails/pull/17743)) ### Notable changes * The `*_filter` family of methods have been removed from the documentation. Their usage is discouraged in favor of the `*_action` family of methods: ``` after_filter => after_action append_after_filter => append_after_action append_around_filter => append_around_action append_before_filter => append_before_action around_filter => around_action before_filter => before_action prepend_after_filter => prepend_after_action prepend_around_filter => prepend_around_action prepend_before_filter => prepend_before_action skip_after_filter => skip_after_action skip_around_filter => skip_around_action skip_before_filter => skip_before_action skip_filter => skip_action_callback ``` If your application currently depends on these methods, you should use the replacement `*_action` methods instead. These methods will be deprecated in the future and will eventually be removed from Rails. (Commit [1](https://github.com/rails/rails/commit/6c5f43bab8206747a8591435b2aa0ff7051ad3de), [2](https://github.com/rails/rails/commit/489a8f2a44dc9cea09154ee1ee2557d1f037c7d4)) * `render nothing: true` or rendering a `nil` body no longer add a single space padding to the response body. ([Pull Request](https://github.com/rails/rails/pull/14883)) * Rails now automatically includes the template's digest in ETags. ([Pull Request](https://github.com/rails/rails/pull/16527)) * Segments that are passed into URL helpers are now automatically escaped. ([Commit](https://github.com/rails/rails/commit/5460591f0226a9d248b7b4f89186bd5553e7768f)) * Introduced the `always_permitted_parameters` option to configure which parameters are permitted globally. The default value of this configuration is `['controller', 'action']`. ([Pull Request](https://github.com/rails/rails/pull/15933)) * Added the HTTP method `MKCALENDAR` from [RFC 4791](https://tools.ietf.org/html/rfc4791). ([Pull Request](https://github.com/rails/rails/pull/15121)) * `*_fragment.action_controller` notifications now include the controller and action name in the payload. ([Pull Request](https://github.com/rails/rails/pull/14137)) * Improved the Routing Error page with fuzzy matching for route search. ([Pull Request](https://github.com/rails/rails/pull/14619)) * Added an option to disable logging of CSRF failures. ([Pull Request](https://github.com/rails/rails/pull/14280)) * When the Rails server is set to serve static assets, gzip assets will now be served if the client supports it and a pre-generated gzip file (`.gz`) is on disk. By default the asset pipeline generates `.gz` files for all compressible assets. Serving gzip files minimizes data transfer and speeds up asset requests. Always [use a CDN](http://guides.rubyonrails.org/asset_pipeline.html#cdns) if you are serving assets from your Rails server in production. ([Pull Request](https://github.com/rails/rails/pull/16466)) * When calling the `process` helpers in an integration test the path needs to have a leading slash. Previously you could omit it but that was a byproduct of the implementation and not an intentional feature, e.g.: ```ruby test "list all posts" do get "/posts" assert_response :success end ``` Action View ----------- Please refer to the [Changelog][action-view] for detailed changes. ### Deprecations * Deprecated `AbstractController::Base.parent_prefixes`. Override `AbstractController::Base.local_prefixes` when you want to change where to find views. ([Pull Request](https://github.com/rails/rails/pull/15026)) * Deprecated `ActionView::Digestor#digest(name, format, finder, options = {})`. Arguments should be passed as a hash instead. ([Pull Request](https://github.com/rails/rails/pull/14243)) ### Notable changes * `render "foo/bar"` now expands to `render template: "foo/bar"` instead of `render file: "foo/bar"`. ([Pull Request](https://github.com/rails/rails/pull/16888)) * The form helpers no longer generate a `