rails--rails/actionpack/CHANGELOG.md

• The +helper+ class method for controllers loads helper modules specified as strings/symbols with String#constantize instead of require_dependency.

  Xavier Noria, Jean Boussier

• Correctly identify the entire localhost IPv4 range as trusted proxy.

  Nick Soracco

• url_for will now use "https://" as the default protocol when Rails.application.config.force_ssl is set to true.

  Jonathan Hefner

• Accept and default to base64_urlsafe CSRF tokens.

  Base64 strict-encoded CSRF tokens are not inherently websafe, which makes them difficult to deal with. For example, the common practice of sending the CSRF token to a browser in a client-readable cookie does not work properly out of the box: the value has to be url-encoded and decoded to survive transport.

  Now, we generate Base64 urlsafe-encoded CSRF tokens, which are inherently safe to transport. Validation accepts both urlsafe tokens, and strict-encoded tokens for backwards compatibility.

  Scott Blum

• Support rolling deploys for cookie serialization/encryption changes.

  In a distributed configuration like rolling update, users may observe both old and new instances during deployment. Users may be served by a new instance and then by an old instance.

  That means when the server changes cookies_serializer from :marshal to :hybrid or the server changes use_authenticated_cookie_encryption from false to true, users may lose their sessions if they access the server during deployment.

  We added fallbacks to downgrade the cookie format when necessary during deployment, ensuring compatibility on both old and new instances.

  Masaki Hara

• ActionDispatch::Request.remote_ip has ip address even when all sites are trusted.

  Before, if all X-Forwarded-For sites were trusted, the remote_ip would default to 127.0.0.1. Now, the furthest proxy site is used. e.g.: It now gives an ip address when using curl from the load balancer.

  Keenan Brock

• Fix possible information leak / session hijacking vulnerability.

  The ActionDispatch::Session::MemcacheStore is still vulnerable given it requires the gem dalli to be updated as well.

  CVE-2019-16782.

• Include child session assertion count in ActionDispatch::IntegrationTest.

  IntegrationTest#open_session uses dup to create the new session, which meant it had its own copy of @assertions. This prevented the assertions from being correctly counted and reported.

  Child sessions now have their attr_accessor overridden to delegate to the root session.

  Fixes #32142.

  Sam Bostock

• Add SameSite protection to every written cookie.

  Enabling SameSite cookie protection is an addition to CSRF protection, where cookies won't be sent by browsers in cross-site POST requests when set to :lax.

  :strict disables cookies being sent in cross-site GET or POST requests.

  Passing :none disables this protection and is the same as previous versions albeit a ; SameSite=None is appended to the cookie.

  See upgrade instructions in config/initializers/new_framework_defaults_6_1.rb.

  More info here

  NB: Technically already possible as Rack supports SameSite protection, this is to ensure it's applied to all cookies

  Cédric Fabianski

• Bring back the feature that allows loading external route files from the router.

  This feature existed back in 2012 but got reverted with the incentive that https://github.com/rails/routing_concerns was a better approach. Turned out that this wasn't fully the case and loading external route files from the router can be helpful for applications with a really large set of routes. Without this feature, application needs to implement routes reloading themselves and it's not straightforward.

  # config/routes.rb
  
  Rails.application.routes.draw do
    draw(:admin)
  end
  
  # config/routes/admin.rb
  
  get :foo, to: 'foo#bar'
  

  Yehuda Katz, Edouard Chin

• Fix system test driver option initialization for non-headless browsers.

  glaszig

• redirect_to.action_controller notifications now include the ActionDispatch::Request in their payloads as :request.

  Austin Story

• respond_to#any no longer returns a response's Content-Type based on the request format but based on the block given.

  Example:

    def my_action
      respond_to do |format|
        format.any { render(json: { foo: 'bar' }) }
      end
    end
  
    get('my_action.csv')
  

  The previous behaviour was to respond with a text/csv Content-Type which is inaccurate since a JSON response is being rendered.

  Now it correctly returns a application/json Content-Type.

  Edouard Chin

• Replaces (back)slashes in failure screenshot image paths with dashes.

  If a failed test case contained a slash or a backslash, a screenshot would be created in a nested directory, causing issues with tmp:clear.

  Damir Zekic

• Add params.member? to mimic Hash behavior.

  Younes Serraj

• process_action.action_controller notifications now include the following in their payloads:

  • :request - the ActionDispatch::Request
  • :response - the ActionDispatch::Response

  George Claghorn

• Updated ActionDispatch::Request.remote_ip setter to clear set the instance remote_ip to nil before setting the header that the value is derived from.

  Fixes #37383.

  Norm Provost

• ActionController::Base.log_at allows setting a different log level per request.

  # Use the debug level if a particular cookie is set.
  class ApplicationController < ActionController::Base
    log_at :debug, if: -> { cookies[:debug] }
  end
  

  George Claghorn

• Allow system test screen shots to be taken more than once in a test by prefixing the file name with an incrementing counter.

  Add an environment variable RAILS_SYSTEM_TESTING_SCREENSHOT_HTML to enable saving of HTML during a screenshot in addition to the image. This uses the same image name, with the extension replaced with .html

  Tom Fakes

• Add Vary: Accept header when using Accept header for response.

  For some requests like /users/1, Rails uses requests' Accept header to determine what to return. And if we don't add Vary in the response header, browsers might accidentally cache different types of content, which would cause issues: e.g. javascript got displayed instead of html content. This PR fixes these issues by adding Vary: Accept in these types of requests. For more detailed problem description, please read:

  https://github.com/rails/rails/pull/36213

  Fixes #25842.

  Stan Lo

• Fix IntegrationTest follow_redirect! to follow redirection using the same HTTP verb when following a 307 redirection.

  Edouard Chin

• System tests require Capybara 3.26 or newer.

  George Claghorn

• Reduced log noise handling ActionController::RoutingErrors.

  Alberto Fernández-Capel

• Add DSL for configuring HTTP Feature Policy.

  This new DSL provides a way to configure an HTTP Feature Policy at a global or per-controller level. Full details of HTTP Feature Policy specification and guidelines can be found at MDN:

  https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Feature-Policy

  Example global policy:

  Rails.application.config.feature_policy do |f|
    f.camera      :none
    f.gyroscope   :none
    f.microphone  :none
    f.usb         :none
    f.fullscreen  :self
    f.payment     :self, "https://secure.example.com"
  end
  

  Example controller level policy:

  class PagesController < ApplicationController
    feature_policy do |p|
      p.geolocation "https://example.com"
    end
  end
  

  Jacob Bednarz

• Add the ability to set the CSP nonce only to the specified directives.

  Fixes #35137.

  Yuji Yaginuma

• Keep part when scope option has value.

  When a route was defined within an optional scope, if that route didn't take parameters the scope was lost when using path helpers. This commit ensures scope is kept both when the route takes parameters or when it doesn't.

  Fixes #33219.

  Alberto Almagro

• Added deep_transform_keys and deep_transform_keys! methods to ActionController::Parameters.

  Gustavo Gutierrez

• Calling ActionController::Parameters#transform_keys/! without a block now returns an enumerator for the parameters instead of the underlying hash.

  Eugene Kenny

• Fix strong parameters blocks all attributes even when only some keys are invalid (non-numerical). It should only block invalid key's values instead.

  Stan Lo

Please check 6-0-stable for previous changes.