Running with Ruby

Tag: Rails 4 (page 1 of 7)

Upgrading to Ruby on Rails 5.0 from Rails 4.2 – application use case

In this article, I’ll try to cover all the issues that I had, when I was upgrading one of my Rails app from Ruby on Rails 4.2 to Ruby on Rails 5.0.

Note: This is not a complete, covering every possible case tutorial. You might encounter issues that I didn’t. If so, feel free to comment and I will try to update the post with other issues and solutions.

Preparations – What should we do before upgrading to Rails 5?

  • Upgrade your Ruby version at least to 2.2.2 (I would recommend 2.3, since it is going to be released soon)
  • Upgrade bundler
  • Upgrade your application to the most recent Rails 4.2 version
  • Check gems compatibility (you may want to be on edge with few gems (or use Rails 5 branches if available))
  • Write more tests if you don’t have a decent code coverage

The last point is the most important. If you don’t have a good code coverage level and you lack tests, upgrading from Rails 4.2 to Rails 5 might be a big problem.

Upgrading to Rails 5 – Gemfile

Currently there are some gems incompatibilities (I will cover them later) and with some, you need to go edge. Here’s a list of gems that I had to switch to edge (or use a specific version) in order to get things running:

gem 'rails', '5.0.0.beta1'
gem 'devise', github: 'plataformatec/devise'
gem 'responders', github: 'plataformatec/responders'
gem 'ransack', github: 'activerecord-hackery/ransack'
gem 'kaminari', github: 'amatsuda/kaminari'
gem 'mailboxer', github: 'mailboxer/mailboxer'

group :test do
  gem 'rspec', github: 'rspec/rspec'
  gem 'rspec-mocks', github: 'rspec/rspec-mocks'
  gem 'rspec-expectations', github: 'rspec/rspec-expectations'
  gem 'rspec-support', github: 'rspec/rspec-support'
  gem 'rspec-core', github: 'rspec/rspec-core'
  gem 'rspec-rails', github: 'rspec/rspec-rails', branch: 'rails-5-support-patches'
  gem 'rails-controller-testing'

Note that I’ve added a rails-controller-testing gem.This gem brings back assigns to your controller tests as well as assert_template to both controller and integration tests. If you use those methods, after upgrading to Ruby on Rails 5, you will have to add it.

After updating your Gemfile you can do a

bundle install

and hopefully you’re ready for the upgrade!

Configuration files – rake rails:update

There are some changes in configuration files (that I will cover), but I would strongly recommend running:

rake rails:update

allow it to overwrite your files and then just add the stuff that was gone (git diff). IMHO it is way less complex approach, that trying to add all new config options manually.

In my case, following files were affected:

On branch rails5
Changes not staged for commit:
    modified:   bin/rails
    modified:   bin/rake
    modified:   config/application.rb
    modified:   config/boot.rb
    modified:   config/environments/development.rb
    modified:   config/environments/production.rb
    modified:   config/environments/test.rb
    modified:   config/initializers/assets.rb
    modified:   config/initializers/cookies_serializer.rb
    modified:   config/initializers/session_store.rb
    modified:   config/initializers/wrap_parameters.rb

Untracked files:

However not all the changes are worth being mentioned (sometimes there were just comments changes).

bin/rails and bin/rake

If you didn’t have any custom stuff there, just go with the flow ;)

config/application.rb and config/boot.rb

I’ll just quote the comment from application.rb (although it is not a new feature it is still worth reminding):

Settings in config/environments/* take precedence over those specified here. Application configuration should go into files in config/initializers all .rb files in that directory are automatically loaded.

And this is exactly what you should do. If you have any custom stuff in application.rb – just move it to initializers. And what about boot.rb? Leave it as it is.


development.rb – feature toggling and file watcher

There’s something quite interesting in new development.rb: feature toggling. Thanks to some files that are in tmp/ you can disable/enable some features to test them in the dev mode, without having to switch to production mode (for example caching).


There is also (commented) file watcher added. You can use it to have an auto-reloading based on your file changes:

# Use an evented file watcher to asynchronously detect changes in source code,
# routes, locales, etc. This feature depends on the listen gem.
# config.file_watcher = ActiveSupport::EventedFileUpdateChecker
test.rb – random order

Tests now will run in a random order. If your tests rely on their execution order, they it is time to rethink them.

config.active_support.test_order = :random
production.rb – no more serve_static_files and Action Cable

Rack cache is out (config.action_dispatch.rack_cache). The config.serve_static_files = false flag is being replaced by:

config.public_file_server.enabled = ENV['RAILS_SERVE_STATIC_FILES'].present?

There are also few configuration options for Action Cable available:

# Action Cable endpoint configuration
config.action_cable.url = 'wss://'
config.action_cable.allowed_request_origins = [ '', /http:\/\/example.*/ ]



New initializer configuration for application controller renderer.  This is a really good feature that finally allows to drop some “bypass like” solutions. If you didn’t render outside of controllers scope, you can leave this commented.

http_host: '',
https: false


Here are all the settings related to CORS. Handle Cross-Origin Resource Sharing (CORS) in order to accept cross-origin AJAX requests. If you don’t know what that means, it means that you can remove this file.


Redis settings for ActionCable. Nothing special except one thing: why do we have config/database.yml and config/redis/cable.yml instead of config/databases/redis.yml and config/databases/mysql.yml (or something similar)?

Starting your freshly updated Rails 5 application

As I said before, if you have a decent code coverage level and you know what you’re doing, upgrade should not be a big problem.

At this point, if you’re not using any fancy route settings or any other magic, you should be able to at least start your application:

rails s

Puma 2.15.3 starting...
* Min threads: 0, max threads: 16
* Environment: development
* Listening on tcp://localhost:3000

However don’t expect it to work correctly… yet ;)

Monkey patching gems

We’re just before Christmas, so I don’t expect gem maintainers to fix them. That’s why we will fix some of them on our own. Here’s a list of gems with some incompatibilities (even when using edge) that I’ve encountered (not including Rails gems):


If you see this error:

Generating an URL from non sanitized request parameters is insecure!

Just use the master branch source code from Github:

gem 'kaminari', github: 'amatsuda/kaminari'

ActiveRecord Import – undefined method type_cast

/activerecord-import/import.rb:479→ block (2 levels) in values_sql_for_columns_and_attributes
/activerecord-import/import.rb:469→ each
/activerecord-import/import.rb:469→ each_with_index
/activerecord-import/import.rb:469→ each
/activerecord-import/import.rb:469→ map
/activerecord-import/import.rb:469→ block in values_sql_for_columns_and_attributes
/activerecord-import/import.rb:468→ map
/activerecord-import/import.rb:468→ values_sql_for_columns_and_attributes
/activerecord-import/import.rb:395→ import_without_validations_or_callbacks
/activerecord-import/import.rb:362→ import_with_validations
/activerecord-import/import.rb:304→ import_helper
/activerecord-import/import.rb:246→ import

To fix this, create an activerecord_import_patch.rb file in initializers and place this code:

class ActiveRecord::Base
  class << self
    def values_sql_for_columns_and_attributes(columns, array_of_attributes)   # :nodoc:
      connection_memo = connection do |arr|
        my_values = do |val,j|
          column = columns[j]

          if val.nil? && == primary_key && !sequence_name.blank?
          elsif column
            if column.respond_to?(:type_cast_from_user)
              connection_memo.quote(column.type_cast_from_user(val), column)
            elsif column.respond_to?(:type_cast)
              connection_memo.quote(column.type_cast(val), column)

ActsAsTaggableOn – Invalid Rails version detection

If you’ll try using ActsAsTaggableOn with Rails 5, you’ll end up with this exception

ArgumentError (Unknown key: :order. Valid keys are: :class_name, :anonymous_class,
:foreign_key, :validate, :autosave, :table_name, :before_add, :after_add, :before_remove,
:after_remove, :extend, :primary_key, :dependent, :as, :through, :source, :source_type,
:inverse_of, :counter_cache, :join_table, :foreign_type, :index_errors):
  app/models/concerns/taggable.rb:6:in `block in <module:Taggable>'
  app/models/text.rb:6:in `include'
  app/models/text.rb:6:in `<class:Text>'
  app/models/text.rb:2:in `<top (required)>'
  app/services/portal/main_service.rb:5:in `<class:MainService>'
  app/services/portal/main_service.rb:3:in `<module:Portal>'
  app/services/portal/main_service.rb:1:in `<top (required)>'
  app/controllers/portal/main_controller.rb:5:in `block in <class:MainController>'
  app/controllers/portal/main_controller.rb:6:in `block in <class:MainController>'
  app/controllers/portal/main_controller.rb:13:in `index'
  lib/middlewares/nofollow_anchors.rb:23:in `call'

it’s directly related to this code:

def has_many_with_taggable_compatibility(name, options = {}, &extention)
  if ActsAsTaggableOn::Utils.active_record4?
    scope, opts = build_taggable_scope_and_options(options)
    has_many(name, scope, opts, &extention)
    has_many(name, options, &extention)

and as you’ve probably already guested, #active_record4? returns false and it will try to use Rails 3 method call. To fix this, create an acts_as_taggable_on.rb file in config/initializers and place there following code:

module ActsAsTaggableOn::Utils
  def self.active_record4?

after a restart, this problem should be solved.

Decent Exposure – methods that no longer exist

This should be considered as a temporary patch. It is not a solid solution (I just wanted to make it work – not to fix decent exposure).

decent_exposure/expose.rb:14:in `block in extended': undefined method `hide_action'
  for ActionController::Base:Class (NoMethodError)
  from /gems/decent_exposure-2.3.2/lib/decent_exposure/expose.rb:7:in `class_eval'
  from /gems/decent_exposure-2.3.2/lib/decent_exposure/expose.rb:7:in `extended'
  from /gems/decent_exposure-2.3.2/lib/decent_exposure.rb:5:in `extend'
  from /gems/decent_exposure-2.3.2/lib/decent_exposure.rb:5:in `block in <top (required)>'
  from /gems/activesupport-5.0.0.beta1/lib/active_support/lazy_load_hooks.rb:38:in `instance_eval'
  from /gems/activesupport-5.0.0.beta1/lib/active_support/lazy_load_hooks.rb:38:in `execute_hook'
  from /gems/activesupport-5.0.0.beta1/lib/active_support/lazy_load_hooks.rb:45:in `block in run_load_hooks'
  from /gems/activesupport-5.0.0.beta1/lib/active_support/lazy_load_hooks.rb:44:in `each'
  from /gems/activesupport-5.0.0.beta1/lib/active_support/lazy_load_hooks.rb:44:in `run_load_hooks'
  from /gems/actionpack-5.0.0.beta1/lib/action_controller/base.rb:262:in `<class:Base>'
  from /gems/actionpack-5.0.0.beta1/lib/action_controller/base.rb:164:in `<module:ActionController>'
  from /gems/actionpack-5.0.0.beta1/lib/action_controller/base.rb:5:in `<top (required)>'
  from /bundler/gems/ransack-2a3759317a44/lib/ransack.rb:38:in `<top (required)>'
  from /gems/bundler-1.10.6/lib/bundler/runtime.rb:76:in `require'
  from /gems/bundler-1.10.6/lib/bundler/runtime.rb:76:in `block (2 levels) in require'
  from /gems/bundler-1.10.6/lib/bundler/runtime.rb:72:in `each'
  from /gems/bundler-1.10.6/lib/bundler/runtime.rb:72:in `block in require'
  from /gems/bundler-1.10.6/lib/bundler/runtime.rb:61:in `each'
  from /gems/bundler-1.10.6/lib/bundler/runtime.rb:61:in `require'
  from /gems/bundler-1.10.6/lib/bundler.rb:134:in `require'
  from /home/mencio/Software/Senpuu/config/application.rb:7:in `<top (required)>

Decent Exposure gem uses some methods from ActionController::Base, that apparently have changed (they no longer exist). To fix that, we will just create dummy methods that will just play along with the Decent Exposure internal logic (config/initializers/decent_exposure.rb):

# Patches for Rails 5 and decent exposure
# Put the line below above the require rails line!
# require File.expand_path('../initializers/decent_exposure', __FILE__)
# require 'rails/all'
module DecentExposure
  module Expose
      def hide_action(action)

    def protected_instance_variables

If it goes about DecentExposure, we need to make one more change. We need to add a require on top of the application.rb file:

# The first and last line are already there - put the decent_exposure one in the middle
require File.expand_path('../boot', __FILE__)
require File.expand_path('../initializers/decent_exposure', __FILE__)
require 'rails/all'

We had to patch this gem before it is loaded, because on load it already performs some logic that requires non-existing methods.

Code base changes

If you’re not planning to use any new Rails features and just (for now) stay with what you had, you won’t have many things to change.

ApplicationRecord instead of ActiveRecord::Base

It is like an ApplicationController. Just an extra inheritance layer just in case you want to have some global (for all models) elements. Just add application_model.rb and change inheritance in all the models:

# Base AR class for all other
class ApplicationRecord < ActiveRecord::Base
  self.abstract_class = true
class Comment < ApplicationRecord

Note: this is not a must be. Your app should work even if you don’t change it.

ActionController::Parameters: params are no longer a Hash

Values that are being sent to params are no longer a hash internally. Now they are a XD, so type checking like this:


have to be replaced with:


ActiveRecord::ReadOnlyRecord: Picture is marked as readonly

Note: this is a dirty hack.

Sometimes I load objects with include and join in a default scope:

scope :with_picture, -> { joins(:picture).includes(:picture) }

default_scope lambda {
  self.require_picture? ? with_picture : where(false)

up until now I could update/remove the main object and its pictures. However now when I fetch object, its pictures are marked as readonly so I get this exception:

ActiveRecord::ReadOnlyRecord: Picture is marked as readonly

activerecord-5.0.0.beta1/lib/active_record/persistence.rb:532→ create_or_update
activerecord-5.0.0.beta1/lib/active_record/callbacks.rb:298→ block in create_or_update
activesupport-5.0.0.beta1/lib/active_support/callbacks.rb:126→ call
activesupport-5.0.0.beta1/lib/active_support/callbacks.rb:506→ block (2 levels) in compile
activesupport-5.0.0.beta1/lib/active_support/callbacks.rb:455→ call
activesupport-5.0.0.beta1/lib/active_support/callbacks.rb:101→ __run_callbacks__
activesupport-5.0.0.beta1/lib/active_support/callbacks.rb:750→ _run_save_callbacks
activerecord-5.0.0.beta1/lib/active_record/callbacks.rb:298→ create_or_update
activerecord-5.0.0.beta1/lib/active_record/suppressor.rb:41→ create_or_update
activerecord-5.0.0.beta1/lib/active_record/persistence.rb:125→ save
activerecord-5.0.0.beta1/lib/active_record/validations.rb:44→ save
activerecord-5.0.0.beta1/lib/active_record/attribute_methods/dirty.rb:22→ save

I wouldn’t be surprised at all (I know this error) but for some reason it was working with 4.2. So if you want to migrate and deal with it later, add this to your dependent models (those for which this error happens):

before_save do
  instance_variable_set(:@readonly, false)

before_destroy do
  instance_variable_set(:@readonly, false)

ActiveRecord::Base default_scope lambda change

Note: I think this also applies to standard scopes.

Not sure if it was a bug or a feature – but it is gone. You can no longer return a nil from a lambda default_scope:

# Won't work anymore
default_scope lambda {
  self.require_file? ? with_file : nil

if you try using it and it will return nil, you will see:

ArgumentError (invalid argument: nil.):
  app/services/portal/main_service.rb:34:in `recent_texts_pictures'
  app/services/portal/main_service.rb:73:in `max_pictures'
  app/services/portal/main_service.rb:46:in `pictures'
  app/services/portal/main_service.rb:19:in `pictures_page'
  app/controllers/portal/main_controller.rb:6:in `block in <class:MainController>'
  app/controllers/portal/main_controller.rb:13:in `index'
  lib/middlewares/nofollow_anchors.rb:23:in `call'

instead of returning nil, just return where(false):

default_scope lambda {
  self.require_file? ? with_file : where(false)

RSpec changes

get/post/put/delete with format also for html

When I don’t provide format: :html, it stopped assuming that is is a html request:

before { get :edit }
before { post :create }
before { put :update }
before { delete :destroy }

examples above fail with following message:

Failure/Error: super(resources, options)

to fix that, just add format: :html to your requestes:

before { get :edit, format: :html }
before { post :create, format: :html }
before { put :update, format: :html }
before { delete :destroy, format: :html }

Note: This seems to be related to responders gem, however adding format fixes the issue.

Symbols to strings for some specs

When matching and expecting resources to receive request parameters, you must send them in string format:

# The first one was no longer working because params now convert it to strings
- let(:mailboxer_message_params) { { mailboxer_message: { body: '' } } }
+ let(:mailboxer_message_params) { { 'mailboxer_message' => { 'body' => '' } } }
before/after actions except/only are now blocks internally

I hade some custom extensions for RSpec, that are checking permissions. Up until now, there were (most of the times) symbols inside:

except: %i( index show )

If we wanted to get it back from the controller (outside of requests scopes) we were doing something like that:

before_actions = subject
  .select { |f| f.kind == :before }

unlesses = { |bf| bf.instance_variable_get(:'@unless') }

this part hasn’t change. We still can obtain our unless/only that way. What did change is what is underneath (in what we get).

In Rails 4.2 we would get strings that we could compare with what we’ve expected:

expect(unlesses.flatten.compact.join).to include "action_name == '#{action}'"

Now each element is a block that accepts a stringified action name and responds with a boolean:

unlesses.flatten.compact.each do |callable|
  #'index') #=> true, etc
  expect( action.to_s))).to eq true
RSpec config settings that are no longer valid

Remove all that I’ve listed below since they are no longer valid:

  • config.disable_monkey_patching!
  • expectations.include_chain_clauses_in_custom_matcher_descriptions = true
  • mocks.verify_partial_doubles = true


As I’ve stated before: If you have a decent code coverage level and you know what you’re doing, upgrade should not be a big problem. However, I still recommend waiting at least until all the most popular gems are up2date.

Good luck!

ActiveRecord count vs length vs size and what will happen if you use it the way you shouldn’t

One of the most common and most deadly errors you can make: using length instead of count. You can repeat this multiple times, but you will always find someone who’ll use it the way it shouldn’t be used.

So, first just to make it clear:

#count – collection.count

  • Counts number of elements using SQL query (SELECT COUNT(*) FROM…)
  • #count result is not stored internally during object life cycle, which means, that each time we invoke this method, SQL query is performed again
  • count is really fast comparing to length
2.1.2 :048 > collection = User.all; nil
 => nil
2.1.2 :049 > collection.count
   (0.7ms)  SELECT COUNT(*) FROM `users`
 => 16053
2.1.2 :050 > collection.count
 => 16053

#length – collection.length

  • Returns length of a collecion without performing additional queries… as long as collection is loaded
  • When we have lazy loaded collection, length will load whole colletion into memory and then will return length of it
  • Might use all of your memory when used in a bad way
  • Really fast when having a eagerly loaded collection
2.1.2 :055 > collection = User.all; nil
 => nil
2.1.2 :056 > collection.length
  User Load (122.9ms)  SELECT `users`.* FROM `users`
 => 16053
2.1.2 :057 > collection = User.all; nil
 => nil
2.1.2 :058 > collection.to_a; nil
  User Load (140.9ms)  SELECT `users`.* FROM `users`
 => nil
2.1.2 :059 > collection.length
 => 16053
2.1.2 :060 > collection.length
 => 16053

#size – collection.size

  • Combines abilities of both previous methods;
  • If collection is loaded, will count it’s elements (no additional query)
  • If collection is not loaded, will perform additional query
2.1.2 :034 > collection = User.all; nil
 => nil 
2.1.2 :035 > collection.count
   (0.3ms)  SELECT COUNT(*) FROM `users`
 => 16053 
2.1.2 :036 > collection.count
   (0.3ms)  SELECT COUNT(*) FROM `users`
 => 16053 
2.1.2 :037 > collection.size
   (0.2ms)  SELECT COUNT(*) FROM `users`
 => 16053 
2.1.2 :038 > collection.to_a; nil
  User Load (64.2ms)  SELECT `users`.* FROM `users`
 => nil 
2.1.2 :039 > collection.size
 => 16053 

Why would you even care?

Well it might have a huge impact on your apps performance (and resource consumption). In general if you don’t want to care at all and you want to delegate this responsibility to someone else, use #size. If you want to care, then play with it and understand how it works, otherwise you might end up doing something like this:

print "We have #{User.all.length} users!"

And this is the performance difference on my computer (with only 16k users):

       user     system      total        real
count     0.010000   0.000000   0.010000 (  0.002989)
length    0.730000   0.060000   0.790000 (  0.846671)

Nearly 1 second to perform such simple task. And this could have a serious impact on your web app! Keep that in mind.


Copyright © 2017 Running with Ruby

Theme by Anders NorenUp ↑