Nested forms If you want to use nested forms like question and answers, good approach is to create! on new action and redirect to edit. That way you have question_id. For new questions or delete questions, you can simply use ajax. So start with rails g scaffold questions title;rails g model answers question:references.

# models/question.rb
class Question < ActiveRecord::Base
  has_many :answers, dependent: :destroy
  accepts_nested_attributes_for :answers, allow_destroy: true

# questions/_form.html.erb
  <div id="answers">
    <% @question.answers.each do |answer| %>
      <%= render partial: 'answer', locals: { question_form: f, answer: answer } %>
    <% end %>
  <%= link_to "Create new answers", create_answer_question_path, remote: true,
  method: :post %>

# questions/_answer.html.erb
  question_form - we need this because we don't want to generate <form> tags
                - we need just fields
  answer - target answer

  we hard code "answers_attributes[]" because
  when we use fields_for :answer, than when we use ajax `new` twice we got two
  and only latest will be considered
  probably because uniq number is reset for each fields_for
  (here id is passed with hidden field)
<%= question_form.fields_for "answers_attributes[]", answer do |ff| %>
  <div class="field">
    <%= ff.hidden_field :id %>
    <%= ff.text_field :content, placeholder: "Answer" %>
    <%= ff.number_field :score, placeholder: 'Score' %>
    <%= ff.number_field :position, placeholder: 'Position' %>
    <%= link_to "Destroy", destroy_answer_question_path(answer.question, answer_id:, remote: true, method: :delete %>
<% end %>

# questions/create_answer.js.erb
<% output = nil %>
<% form_for(@question) do |f| %>
  <% output = j render partial: 'answer', locals: { question_form: f, answer:
  @answer } %>
  <% end %>
$('#answers').append('<%= output %>');

# config/routes.rb
  resources :questions do
    member do
      post :create_answer
      delete :destroy_answer

# controllers/questions_controller.js
  def create_answer
    @answer = @question.answers.create!

  def destroy_answer
    @answer = @question.answers.find(params[:answer_id])

    def question_params
      params.require(:question).permit(:title, :time_limit, answers_attributes: [:id, :score, :content, :_destroy] )

You can try cocoon gem and use link_to_add_association tutorial post


Some usefull validations, like validate email regexp

validates :email, format: { with: /\A([^@\s]+)@((?:[-a-z0-9]+\.)+[a-z]{2,})\z/i }
validates :email, format: { with: User.email_regexp, allow_blank: true }
validates :email, uniqueness: { scope: :user_id }

When validating associations, always is _id since we use that in select input inside form (don’t validate object validates :job_type, presence: true since error will not be shown on :job_type_id).

validates :job_type_id, presence: true

<%= :job_type_id, { |job| [,] }, {
prompt: true }, class: 'my-class' %>

form select can accept as 2th param (choices) two variant:

as 3th param (options):

as 4th params (html options)

You should avoid saving without validation save(validate: false) or update_attribute :name, 'my name'. It is risky to save without validations. Other methods also do not check validation update_column, @users.update_all.

Conditional validations can be used with proc new like if: -> { } but with parameters. Also if: lambda {|a| } (difference in required params to block)

validates :password, confirmation: true, if: { |a|

Validate occurs :on :save by default (but you can not use validate :d, on: :save). you can split and specify to run on on: :create or on: :update. To run validation on destroy you need hook before_destroy where you can errors.add and return false so hook reverts.


NOTE THAT YOU SHOULD NOT USE HOOKS… Better is to just call a method where needed, because you will not needed it always (for example in tests you want different value) If you really need (default value that you really want) than please use ||= conditional assignment Default value for column could be in migration but than you need another migration if you want to change value. If we put on after_initialize :default_values_on_initialize than it is called also when you read object (that is required if you want to change default values for existing objects) If you have validates :logo than you can not put on before_save :default_logo since validation will fail before that. We need before_validation which occurs only on update and create.

Note that business logic could be that some fields could be cleared. For example some logo is default value, but someone could completely remove logo. So you need to separate default_values_on_create and default_values_on_update (which could get nil on update some fields)

after_initialize :default_values_on_initialize
before_validation :default_values_on_create, on: :create
before_validation :default_values_on_update, on: :update


def default_values_on_initialize
  self.some_no_nil_number ||= 1
  self.some_positive_number = 1 unless some_positive_number.to_i > 0
  self.some_false_value ||= false
  # do not use self.some_true_value ||= true since that will override if
  # some_true_value = false
  self.some_true_value = true if some_true_value.nil?
  # return value should be true or nil

def default_values_on_create
  self.logo ||= Rails.application.secrets.default_restaurant_logo
  # return value should be true or nil

TO calculate some files you can use after_save :update_total. Do not use after_update :update_total since you can not update in that method.

Custom validations validate :my_method or validate { |customer| customer.check_permissions } should add errors to the object (return value is not important and could be false).

class Customer
  def my_method
    errors.add(:name) if name != 'Duke'
  def check_permissions
    errors.add(:name) if name != 'Duke'

Default Order

default_scope { order('created_at DESC') }

Usually default scope is not good practice (except for order). Even to trashable concern is bad usage, for example if you use trashed on user and comments model

rails g migration add_trashed_to_users trashed:boolean
rails g migration add_trashed_to_comments trashed:boolean

# user.rb & comment.rb
  default_scope { where(trashed: nil) }
  scope :trashed, -> { unscoped.where(trashed: true) }

than query User.first.comments.trashed will return all trashed comments from ALL users! unscoped will remove even association scopes, thanks Joseph Ndungu on comment.

Uncope can receive params what to unscope, example User.unscope(:where). If you use this in belongs_to :location, -> { unscope :where } than it will also remove association belongs to condition, so you have to unscoped only columns from default_scope belongs_to :location, -> { unscope where: :operator_id } excellent link how to remove scope I usually create additional classes with self.default_scopes = [] so it does not use default scope. Also set table name so I can use sql.

class UnscopedCustomer < Customer
  self.default_scopes = []
  self.table_name = "customers"
  belongs_to :unscoped_company, foreign_key: :company_id

class UnscopedCompany < Company
  self.default_scopes = []
  self.table_name = "companies"
  has_many :customers, foreign_key: :company_id

Format date

Write datetime in specific my_time format

# config/initializers/mytime_formats.rb
# all 3 classes
# puts user.updated_at.to_s :myapp_time
# puts :myapp_time
# puts :myapp_time # Date need to be type casted to Time
# puts :myapp_date # Time object to Date if we want myapp_date
Time::DATE_FORMATS[:default] = "%d-%b-%Y %I:%M %p" # this is same as for
# datepicker format = 'DD-MMM-YYYY h:mm A'
Time::DATE_FORMATS[:at_time] = lambda { |time| time.strftime("%b %e, %Y @ %l:%M %p") }

Date::DATE_FORMATS[:myapp_date] = lambda { |date| date.strftime("%b %e, %Y") }
Date::DATE_FORMATS[:myapp_date_ordinalize] = lambda { |date| date.strftime("#{} %b %Y") } and are using system time. If you are using browser-timezone-rails than timezone will be set for each request. But if you need something from rails console, you need to set timezone manually. = 'Belgrade' (list all in rake time:zones:all). It is good to always use and Rails helpres use zone (

Change system timezone (which is used by browsers) with sudo dpkg-reconfigure tzdata. Note that rails c uses UTC # => "UTC", but byebug in rails s returns default time zone # CEST Europe. You can use users timezone with browser-timezone-rails. will return offset to UTC in seconds. I do not know why returns different results (3600 instead of 7200).

Multiline render js response

Write long string in multiple lines with %(), for example:

format.js do
  render js: %(
    $('##{key}').replaceWith('#{view_context.j view_context.render partial: 'product_table', locals: { products: Product.send( key).all, product_type_string: key.to_s}} ');

If you receive a lot of errors An ActionView::MissingTemplate occurred in and Missing template customer/sessions/new, customer_application/new, application/new with {:locale=>[:in, :en], :formats=>["Application/*"], :variants=>[], :handlers=>[:erb, :builder, :raw, :ruby, :coffee, :jbuilder]}. Searched in: on some landing or login pages than simply add:

# app/controllers/home_controller.rb
class HomeController < ApplicationController
  respond_to :html

  def index
    respond_with do |format|
      format.html { render :index }
      format.any { render :index }

Another solution to respond to all formats is to add formats argument

# app/controllers/home_controller.rb
class HomeController < ApplicationController
  def index
    render :index, formats: [:html]

So this will render index for any type of requests

curl http://loc:3001
curl http://loc:3001 -H "Accept: application/json"


Textarea should be autosized (download autosize.js from dist/ folder). Put below your textarea for ajax responses and put in your main.js file on jQuery load.

<%# app/views/forms/_form.html.erb %>
<%= f.text_area :title %>
<%= javascript_tag "autosize($('textarea'))" if request.xhr? %>

// app/assets/javascripts/main.js.erb
$(function() {

Upload file

Since default file button does not look nice we just hide it (android has some probles when it is hidden, so we just move it), and use label to trigger it and on change we submit the form. Note that we use id because we could have more than one form on a page.

<%# _suppression_list.html.erb %>
  <div class="pull-right">
    <%= form_tag import_suppression_path(suppression_list), multipart:true, class: 'pull-right' do %>
      <span data-chosen-filename></span> &nbsp;
      <%= label_tag :file, 'Upload CSV', for: "upload-file-#{}", style: 'display:inline' %>
      <%= file_field_tag :file, id: "upload-file-#{}", 'data-upload-file': true, style: 'position:absolute; top: -1000px' %>
    <% end %>
# javascripts/
$(document).on 'change', '[data-upload-file]', (e) ->
  console.log("upload csv")
  file = this.files[0]
  allowedExtensions = /^(text\/csv)$/
  $(this).closest('form').find("[data-chosen-filename]").text this.value
  if (this.files.length == 0 )
    alert("Please chose one csv file")
  if(! allowedExtensions.test(file.type))
    alert("File type is not allowed")


Show all databases in postgresql, first change user to postgres

sudo su -l postgres

# or in one command
sudo -u postgres psql
\connect database_name


Active record Default strategy is :nullify ie keep row, but set post_id = nil. So always use has_many :association, dependent: :destroy so it will not use nullify but delete_all strategy link

Add json and hstore

Note that you can use simple text column (without default value) and add serialize :preferences, Hash in your model, so user.prefereces # => {} is defined for initial empty value.

Adding array of any type and hstore is easy, just add default value [] and '' (breaks for {}). You can create hstore extension in migration. If you need arrays of hstore than you need to go for json or jsonb (better optimization only pq 9.4) or json/jsonb array.

class CreatePhones < ActiveRecord::Migration
  def change
    execute "create extension hstore"
    create_table :phones do |t|
      t.string :my_array, array: true, default: []
      t.hstore :my_hash, default: ''
      t.jsonb :my_json, default: []
      t.jsonb :my_array_of_json, array: true, default: []

DB user need to be superuser to be able to create extension. If you need to run rake db:migrate:reset to rerun all migration to check if db/schema is in sync, than the best way is to alter user to have superuser privil sudo su postgres -c "psql -d postgres -c 'ALTER USER orlovic WITH SUPERUSER;'" where orlovic is database username (you can see those usernames - roles using pqadmin visual program) link.

You can enable hstore manually link sudo psql -d Scuddle_app_development -U orlovic and CREATE EXTENSION hstore; or in one command: sudo su postgres -c "psql Scuddle_app_test -c 'CREATE EXTENSION hstore;'". You should do this also for test and development database. If you don’t know database name you can use rails runner

sudo su postgres -c "psql `rails runner 'puts ActiveRecord::Base.configurations["production"]["database"]'` -c 'CREATE EXTENSION hstore;'"

Dump database

Dump database from production for local inspection, you can download from heroku manually or using commands:

heroku pg:backups:capture # it will create b002.dump
heroku pg:backups:download # it will download to latest.dump

Save it for example tmp/b001.dump.

You can dump LOCAL database with pg_dump. Note that this is plain sql, but heroku dump is binary format (size is much smaller).

# or heroku style, replace: mypassword myuser and mydb
# PGPASSWORD=mypassword pg_dump -Fc --no-acl --no-owner -h localhost -U myuser $DATABASE_NAME > $DUMP_FILE

Restore database

Restore from local textual and binary dump

export DUMP_FILE=tmp/b001.dump
export DATABASE_NAME=$(rails runner 'puts ActiveRecord::Base.configurations["development"]["database"]')
chmod a+r $DUMP_FILE

rake db:drop db:create

# textual dump

# binary dump
sudo su postgres -c "pg_restore -d $DATABASE_NAME --clean --no-acl --no-owner -h localhost $DUMP_FILE"

Or you can use duleorlovic’s load_dump helper

To restore on heroku you need to dump with same flags (dump is binary) and push the file somewhere on internet, for example AWS S3 and than run in console

heroku pg:backups restore --confirm playcityapi DATABASE_URL

Heroku upgrade database plan

Upgrade heroku hobby-dev na hobby-basic ($9/month max 10M rows). All plans

heroku addons:create heroku-postgresql:hobby-basic
# Creating heroku-postgresql:hobby-basic on ⬢ myapp... $9/month
# Database has been created and is available
#  ! This database is empty. If upgrading, you can transfer
#  ! data from another database with pg:copy
# Created postgresql-defined-42601 as HEROKU_POSTGRESQL_CHARCOAL_URL
# Use heroku addons:docs heroku-postgresql to view documentation

# ▸    WARNING: Destructive action
#  ▸    This command will remove all data from CHARCOAL
#  ▸    Data from DATABASE will then be transferred to CHARCOAL
#  ▸    To proceed, type myapp or re-run this command with --confirm myapp
# > myapp
# Starting copy of DATABASE to CHARCOAL... done
# Copying... done


Upgrading from hobby-basic to standard-0

heroku pg:info
heroku addons:create heroku-postgresql:standard-0
# save the variable name HEROKU_POSTGRESQL_{some color}_URL
heroku maintenance:on
heroku pg:promote HEROKU_POSTGRESQL_color_URL
heroku maintenance:off
heroku config:set DATABASE_URL=....url from config
heroku addons:destroy HEROKU_POSTGRESQL_color_old_URL

Postgres tips


if you want to redo migration (revert and migrate again) you can use redo To drop database in console you can ActiveRecord::Migration.drop_table(:users)

Has_many through

has_many :roles; has_many :projects, through: roles can be used for many to many associations. You can automatically add new role, no need to

user.roles.create project: project
# or
user.projects << project

has_and_belongs_to_many (habtm) can be generated with rails g migration create_campaigns_templates campaign:references template:references and add id: false as suggested Note that both model names are plural. But rubocop suggest to use separate model for habtm relation (in that case I use singular first model).

class CreateCampaignsTemplates < ActiveRecord::Migration
  def change
    create_table :campaigns_templates, id: false do |t|
      t.references :campaign, foreign_key: true, null: false
      t.references :template, foreign_key: true, null: false


and adding has_and_belongs_to_many :templates in those classes.

You can force uniq with add_index :campaign_templates, [:campaign_id, :template_id], unique: true and check in rails with

campaign.templates << template unless campaign.templates.include? template
# remove

If you have different name of the table: t.references :donor, foreign_key: true, null: false but donor is actually in users table, than you need to change foreign key (note that add_foreign_key params are in plurals!)

# to_table is in rails5
t.references :donor, foreign_key: { to_table: :users }, null: false
# in rails4 use references keyword also as parameter, not that foreight key
# constrains need to be added in separate command. note suffix "_id"
t.references :donor, foreign_key: false, null: false, references: :users
add_foreign_key :table, :users, column: :donor_id

or manually write relation

create_table :donations
  t.belongs_to :donor, index: true, null: false
add_foregn_key :donations, :users, column: :donor_id

Also in model you need to write: has_many :donations, foreign_key: :donor_id and belongs_to :donor, class_name: "User"

Note that t.references should be used with foreign_key: true, null: false since foreign_key is not automatically used (t.references automatically add index). Note that t.belongs_to by default use index: false, so you need to use index: true (or add_index later). But usually you use add_foreign_key that will also add index (name will be like: fk_rails_123123) if it is not added by belongs_to, so you can safelly use t.belongs_to ... index: false if you are using add_foregn_key.

For MySql I have to use unsigned: true in t.belongs_to :user, null: false, unsigned: true or for t.references :user, unsigned: true You can add column at specific location add_column :feeds, :last_modified, :datetime, after: :url (after column does not work in psql since you need to change table)

If you need to update specific fields of association, add validation or cdd ustom order then you should create the model and use has_many :templates, through: campaign_templates, order: 'campaign_templates.created_at ASC' Note that here we user singular first part so model name looks better

class CampaignTemplate < ActiveRecord::Base
  belongs_to :campaign
  belongs_to :template

If you accidentaly use belongs_to :templates (pluralized) than error is undefined method relation_delegate_class' for Templates:Module

Database indexes

Indexes should be on all columns that are references in WHERE, HAVING, ORDER BY parts of sql. For example if you find using specific column User.find_by column: 'val'. Also we can add index to :updated_at column since we sometimes order by that column.

All foreign keys need to have index. You can use add_reference (adding column and index). if you want to add or remove reference in migration rails g migration add_user_to_leads user:references which will generate

class AddUserToLeads < ActiveRecord::Migration
  def change
    # reference will automatically include index on that column
    add_reference :leads, :contact, foreign_key: true
    assert_difference 'Fetch.count', 1 do
      assert_difference 'TestItem.count', 1 do, parser).perform

For polymorphic associations owner_id and owner_type

class Organization < ActiveRecord::Base
  has_many :projects, :as => :owner

class User < ActiveRecord::Base
  has_many :projects, :as => :owner

class Project < ActiveRecord::Base
  belongs_to :owner, :polymorphic => true

When you create you can use references and polymorphic: true

  t.references :featureable, polymorphic: true, null: false

If you add one by one column than you need to add double index:

t.string :owner_type
t.integer :owner_id

# Bad: This will not improve the lookup speed
add_index :projects, :owner_id
add_index :projects, :owner_type

# Good: This will create the proper index
add_index :projects, [:owner_type, :owner_id]

You can add unique index on some existing columns (to see validation error instead of database esception, should also be in rails validates :email, uniqueness: { scope: :user_id })

class AddUniqContacts < ActiveRecord::Migration
  def change
    add_index :contacts, [:email, :user_id], unique: true

Do not add index for tables that has a lot or removing, since perfomance will be bad. Also huge tables need huge indexes, so pay attention on size.

Remove index and foreign key if you want to remove column

  remove_foreign_key :online_payment_responses, :location_packages
  remove_column :online_payment_responses, :location_package_id
  if index_exists?(:online_payment_responses, name: 'fk_rails_b36eeaa704')
    remove_index :online_payment_responses, name: 'fk_rails_b36eeaa704'

In mysql sometimes LIMIT 10 is slower than LIMIT 100 since it won’t use index for small stuff, but if table is huge than it’s much slower without index. To force index use something like User.from("'users' FORCE INDEX (my_user_index") link If you receive error undefined method map for "'users' FORCE INDEX (my_user_index)":Arel::Nodes::SqlLiteral than you can try to replace includes/references with joins.


For errors

Mysql2::Error: Lock wait timeout exceeded; try restarting transaction
Mysql2::Error: Deadlock found when trying to get lock; try restarting transaction: UPDATE

you can see examples on To get latest deadlock in mysql run msql> SHOW ENGINE INNODB STATUS

Farmer.lock.find will wait untill it is free to lock this farmer.

alert = nil
ActiveRecord::Base.transaction do
  if something_wrong?
    alert = 'Can not ...'
    raise ActiveRecord::Rollback
if alert

Race conditions

When you have a lot of users and long db update commands than you probably have exceptions for deadlocks and duplicate entries. You can retry

# app/controllers/users_controller.rb
  def update
    retries ||= 2
    # perform some long task and in another browser and thread try to lock it
  rescue ActiveRecord::RecordNotUnique
    retries -= 1
    raise unless retries.nonzero?


On Heroku it is better to use JawsDB MySQL than ClearDB MySQL since it has more MB in for free usage.


$(function) is shorthand for $(document).ready(function(){}) (and is not the same as $(document).on('ready', function(){})) and it is when DOM is ready. If you need to know when all images and iframes are loaded than use $(window).on('load', function() {}). Turbolinks is installed with adding gem 'turbolinks' and include in application.js //= require turbolinks.

Turbolinks intercepts all clicks on links to the same domain. When you click an eligible link, Turbolinks prevents the browser from following it. Instead, Turbolinks changes the browser’s URL using the History API, requests the new page using XMLHttpRequest, and then renders the HTML response.

During rendering, Turbolinks replaces the current <body> element outright and merges the contents of the <head> element. The JavaScript window and document objects, and the HTML <html> element, persist from one rendering to the next.

With turbolinks rails acts as single page application. It will intercept <a> links, but if you want to redirect using javascript window.location than you can use Turbolinks.visit link, for example

$(document).on 'click', '[data-click]', (e) ->
  Turbolinks.visit $(this).data('click')

Each visit can be advance or replace, than can be set with data-turbolinks-action="replace" or Turbolinks.visit link, { action: 'replace' }). restore visit is when we click Back and can not be canceled. You can disable turbolink on specific link with data-turbolinks="false"

Since window and document remains, all objects remain in memory.

So if you want to do something on every page and on first load than you need to bind on turbolinks:load

// app/assets/javascripts/
$(document).on('turbolinks:load', ->
  # in previous rails we used to catch up on 'ready page:load'
  console.log "document on turbolinks:load"
  # Activating Best In Place
  autosize $('textarea')

  $('[data-disable-button-if-empty]').on('change paste keyup input', ->
    disabled = this.value.length == 0
    button = $(this).parents('form').first().find('[type=submit]')
    button.prop('disabled', disabled)

  # initial status
  $.each $('[data-disable-button-if-empty]'), (index, el) ->

  $('[data-on-change-submit]').on 'change', ->

When you are using data-remote-true for show edit form than request is JS and any javascript inside js.erb or inside partial $('#id').replaceWith('<%= j render 'partial' %>'); is executed every time. Problem is with bootstrap modals where request is HTML. remote modal content is deprecated in v4 so it’s better to use custom implementation

# app/assets/javascripts/
$(document).on 'click', '[data-js-modal]', (e) ->
  arget = this.dataset.jsModal
  remote_link = this.href || this.dataset.jsModalHref
  # use $.ajax and replaceWith since $.load use .html which discard scripts
    url: remote_link
  ).done (responseText) ->
    $(target + " .modal-content").replaceWith responseText

and trigger turbolinks load so it can catch new items. Instead of triggering, you can call initializeDatepicker(). That is also needed below the forms since ajax request in remote-true forms also need to run the initializaion scripts.

# app/assets/javascripts/
window.initializeDatepicker = ->
  $date_elements = $('.date')
  return unless $date_elements

If you want to perform something on click for existing and new elements, but only on particular page, you can bind bind click on document on that page inside body. But when you navigate 10 times, it will trigger 10 times. So you can unbind on page:before-change. Note that you should write that after function declaration because when you click on page again, it will assign “on” on previous version, and unassign latest version of your_function

function your_function() {
  LOG && console.log("your_function call");
$(document).on('click', '.class', your_function);
$(document).one('page:before-change',function() {
  LOG && console.log("unassign your_function");
  $(document).off('click', your_function);


If you want indempotent seeds data you should have some identifier (for example id) for wich you can run where(id: id).first_or_create! do...end. User should use first_or_initialize since we can’t create without password, but we don’t have password field. do ... end block is used only if that object is not found. slice is used for uniq fields (not generated by faker), but all other fields should be populated in block. Slice is nice method to get params hash from current active record object user.slice :email, :name # { email: '', name: '' }

Use faker gem to generate example strings:

# db/seeds.rb
# we keep all variables (defined as property var: :name) in global b hash
# so you an access them later as `b[:name]`
b = {}

# rubocop:disable Rails/Output
# JobType
  { var: :my_job_type, name: 'Admin/Office' }
].each do |doc|
  r = JobType.where(doc.except(:var)).first_or_create! do |job_type|
    # you do not need to call save! here
    # put common stuff here
    job_type.domain = 'my_domain'
    puts "JobType #{doc[:name]}"
  b[doc[:var]] = r if doc[:var]

# deterministic and random data
1.upto(10).map do |i|
  { email: "user#{i}@asd.asd", password: Faker::Internet.password }
end.each do |doc|
  User.where(doc.except(:password, :remote_avatar_url)).first_or_create! do |user|
    user.password = doc[:password]
    user.remote_avatar_url = doc[:remote_avatar_url]
    user.confirm! # do not skip_confirmation!, we need to have confirmed so
    # we can log in as this user from admin panel. sign_in will fail for
    # autogenerated users who did not confirm or skip confirmation
    # note that admin can NOT sign in as any not confirmed user
    puts "User #{}"

asd_user = User.find_by! email: 'asd@asd.asd'

Custom logger

You can add tags (methods for request object), for example:

# config/application.rb
config.log_tags = [:remote_ip]

But if you want custom logger, than you can override existing

# config/initializers/logger_formatter.rb
class ActiveSupport::Logger::SimpleFormatter
  # from activesupport/lib/active_support/core_ext/logger.rb
  def call(severity, time, progname, msg)
    "#{severity_color severity} #{String === msg ? msg : msg.inspect}\n"


  def severity_color(severity)
    case severity
    when "DEBUG"
      "\033[0;34;40m[DEBUG]\033[0m" # blue
    when "INFO"
      "\033[1;37;40m[INFO]\033[0m" # bold white
    when "WARN"
      "\033[1;33;40m[WARNING]\033[0m" # bold yellow
    when "ERROR"
      "\033[1;31;40m[ERROR]\033[0m" # bold red
    when "FATAL"
      "\033[7;31;40m[FATAL]\033[0m" # bold black, red bg
      "[#{severity}]" # none

I need to put backtrace in reverse order, so I wrap that code (for example whole seeds.rb).

# code
# never rescue from Exception, but use Standard error
rescue StandardError => e
  puts e.backtrace.reverse
  puts e.class, e.message

I do not know how to catch all exceptions without wrapping (maybe to use rack app as the exception_notification does). I will try to create new logger, it is used for whole Rails application.


mustache is nice to render user templates

Usage is simple as

data_for_body == current_user.contacts.first
@template_body =, data_for_body).render
# in view use <%= simple_format @template_body %> to convert \n to <br>

It is usefull to delegate fields in contacts model to some belongs_to association, like delegate User::FIELD_NAMES, to: :user.

# app/services/template_render_service.rb
class TemplateRenderService
  attr_reader :template_body, :data_for_body

  def initialize template_body, data_for_body
    @template_body = template_body
    @data_for_body = data_for_body

  def render
    m =
    m.raise_on_context_miss = true
    m.render(template_body, data_for_body)
  rescue Mustache::ContextMiss, Mustache::Parser::SyntaxError => e

Rake tasks

You can write tasks with arguments rails rake

# lib/tasks/db.rake
namespace :db do
  desc 'This task does nothing'
  task nothing: :environment do
    # environment is needed to load rails

Instead of :env or :environment you can use other tasks on which it depends. Also you can receive parameters into args

# lib/tasks/update_subdomain.rake
namespace :update_subdomain do
  desc "update subdomains for my-user. default value is 'my_subdomain'"
  task :my_user, [:subdomain] => :environment do |task, args|
    args.with_defaults subdomain: 'my_subdomain'
    user = User.find_by name: 'my-user'
    fail "Can't find user 'my-user'" unless user
    user.subdomain = args.subdomain!
    puts "Updated #{user.subdomain}" || next if return_from_rake_task_now?
    puts 'Finished'

# run with
rake update_subdomain:my_user
rake update_subdomain:my_user[]
rake update_subdomain:my_user[new_subdomain]

Another use of rake is to seed

# lib/tasks/seed.rake
namespace :seed do
  task :bid_types => :environment do
    ["live", "silent", "teacup"].each do |name|
      BidType.find_or_create_by! name: name

# db/seed.rb

# call task from another task
Rake::Task['translate:copy'].invoke Rails.env

so you can create with rake db:seed or rake seed:bid_types

If you want to know inside some code whether you run from rake or from rails (for example you do not want to send emails for seed users), you can use

if File.basename($0) == "rake"
  # I'm from rake
  # I'm from rails

Sessions and share cookies on multiple subdomains

If you use sharing-cookies-across-subdomains-with-rails-3 or what-does-rails-3-session-store-domain-all-really-do

# config/initializers/session_store.rb
Rails.application.config.session_store :cookie_store, key: '_myapp_session',
domain: :all, expire_after: 60.minutes

This works fine for top level domains and subdomains for them.

For example if you have domain: :all and you sign in at asd.local, than you will be signed in also on asd.asd.local and asd.asd.asd.local and so on… Cookie will be with domain .asd.local for all those sites.

You need to know that some domains like belongs to Public Suffix List so browser will prevent setting domain: :all cookie for them (because someone can use to read cookies from

Note you can not login at (PSL and browser will prevent cookie).

Rails cookie store knows for public suffix list, so for it will use and all others like it will use .in.local.

You can login at but it wont be shared to Cookie will be with domain, so you will be signed in also on

Cookie are send on each request. Cookies usually does not expire, but you can set expire_after: 60.minutes. They are send again when you refresh close & open window.

Note that flash messages are also for each domain. So if you redirect from one domain to another, you can not use flash messages. Old flash message will be shown when user come back to previous domain (on which request flash was set).

Subdomains and long domains

Rails has method extract domains but it requires second param which determine if it is top level and second level domain.

ActionDispatch::Http::URL.extract_domain '', 1 # ''
ActionDispatch::Http::URL.extract_domain '', 2 # ''

You could try to guess based on next to last string and determine if it is shorter or equal than 3 chars. Rails domain is always last two strings.

# = ''
request.domain = ''
request.subdomain = 'dule.asd'

# you can specify different tld length
request.domain(2) = ''

When you use url for, you can pass host parameter. But if you pass also subdomain than host param will be trimmed to only last two strings issue

ActionDispatch::Http::URL.url_for host: ''
ActionDispatch::Http::URL.url_for host: '', subdomain: 'sub'
# note that we are missing 'asd' in domain name

Text syntax on buttons and messages

Use “Register” or “Sign up” for registrations and “Log in” and “Log out” for sessions (since it is easy to differentiate from “sign up”).

7 patterns

From 7-ways-to-decompose-fat-activerecord-models suggestions for code organization I use mostly:

You can check also You can organize domain (controller, model and view) into separate folders drawers

Service objects

Always define only one public method: call, perform or process

# app/services/match_posts.rb
def MatchPosts
  def initialize(posts)
    @posts = posts

  def perform

or more complex with exception rescue.

# app/services/my_service.rb
class MyService
  class Result
    attr_reader :message
    def initialize(message)
      @message = message

    def success?

  class Error < Result
    def success?

  # Some custom exception if needed
  class ProcessException < Exception

  def initialize(h)
    @user = h[:user]

  def process(posts)
    success_message = do_something posts success_message
  rescue ProcessException => e e.message


  def do_something(posts)
    raise ProcessException, "Error: empty posts" unless posts
    "Done with do_something"
# main.rb
require './my_service.rb'

my_service = user: 'me'
puts my_service.process(1).success? # true
puts my_service.process(1).message # Done with do_something
puts my_service.process(false).success? # false
puts my_service.process(false).message # empty posts

Even simpler Result class (not using Error class) but not recomended.

  class Result
    attr_reader :error, :message
    def initialize(error:, message:)
      @error = error
      @message = message
    def success?
  end message: success_message error: e.mesage

Service object is similar to Command pattern which is implemented in gem

Form Objects

form objects (query objects) for multiple in multiple out data

# app/form_objects/landing_signup.rb
class LandingSignup
  include ActiveModel::Model
  FIELDS = %i[ current_city current_location current_group prefered_group email].freeze

  validates :email, presence: true

  def save

  # no need for initialize since AR will pick from params


# app/controllers/pages_controller.rb
class PagesController < ApplicationController
  def home
    @landing_signup =
    @landing_signup.current_city = City.first

  def landing_signup
    @landing_signup = landing_signup_params
    @landing_signup.current_city = City.first
    if @landing_signup.perform
      sign_in @landing_signup.user
      redirect_to dashboard_path, notice: @landing_signup.notice
    else[:alert] = @landing_signup.errors.full_messages.join(', ')
      render :home


Decorators can be used instead of callbacks. Differs from service object because it just decorate existing model (it can be used instead of that model, usefull for forms). Form object is like a decorator, but for multiple models. So instead using method_missing you can use standard ruby SimpleDelegator which delegates any method to object that was passed in initialization

# app/decorators/message_decorator.rb
class MessageDecorator < SimpleDelegator
  def message

  def save_and_send_notifications && _send_notifications

  def _send_notifications do |move|
      next if move.user == message.user

Use in controller

  def show
    @message_decorator =

  def create_message
    @message_decorator = _message_params
    if @message_decorator.save_and_send_notifications
      redirect_to chat_path(@chat), notice: t_crud('success_create', Message)
      render :show


DHH and interesting is trashable video blog example usage

module Someable
  extend ActiveSupport::Concern

  included do
    # define validations, callbacks and associations here (before_ has_  macros)
    # access module variables @@my_module_variable or class variable
    # self.class.class_variable_get :@@someable_value
    has_many :something_else, as: :someable
    class_attribute :tag_limit

  # instance methods are defined here

  # methods defined here are going to extend the class, not the instance of it
  # do not use "self." in method definition
  # you can set "@@my_module_variable" here
  # better is to set class_variable:  self.ancestors.first.class_variable_set :@@someable_value, 'some value'
  # also you can include other concerns here
  class_methods do
    def tag_limit(value)
      klass = self.ancestors.first
      previous_columns = klass.class_variable_get(:@@monetized_columns) if klass.class_variables.include? :@@monetized_columns
      previous_columns ||= []
      klass.class_variable_set :@@monetized_columns, previous_columns + columns
      self.tag_limit_value = value

My implementation of friendly_id gem


We an use observable objects to send notifications implementation in 3 languages It could looks like action as a distance antipattern but if we explicitly add than is it fine.

module ObservableImplementation
  def observers
    @observers ||= []

  def notify_observers(*args)
    observers.each do |observer|

  def add_observer(object)
    observers << object

class Task
  include ObservableImplementation
  attr_accessor :counter
  def initialize
    self.counter = 0

  def tick
    self.counter += 1

class PutsObserver
  def initialize(observable)
    observable.add_observer self

  def update(counter)
    puts "Count has increased by #{counter}"

class DotsObserver
  def initialize(observable)
    observable.add_observer self

  def update(counter)
    puts "." * counter

task =
task.tick # ..
task.tick # ...
# Count has increased by 3

Unsubscribe links

# app/views/layouts/_email_footer.html.erb
  <%= link_to "Unsubscribe", link_for_unsubscribe(user, controller.mapping_to_unsubscribe_group(controller.action_name)) %> from this type of email.
  <%= link_to "Manage", settings_user_url(user) %> which emails you receive.

# app/controllers/application_mail.rb
  def mapping_to_unsubscribe_group method_name
    group = ApplicationHelper::UNSUBSCRIBE_MAPPING_METHOD_TO_GROUP[method_name.to_s]
    if group.nil?
      puts "!!!!! No group found for method_name=#{method_name.to_s}"
      ExceptionNotifier.notify_exception("just to notify that there is no group for method_name #{method_name}") )

# app/helpers/application_helper.rb
    "first_application_instructions" => "tips_and_help_emails_jobseeker",
    "application_in_review_instructions" => "tips_and_help_emails_jobseeker",
    "new_candidate_email" => "new_candidate_or_applicant",
  def link_for_unsubscribe user, unsubscribe_group
    referral_token_and_unsubscribe_group = user.referral_token + unsubscribe_group.to_s
    unsubscribe_url key: Base64.encode64(referral_token_and_unsubscribe_group)

# config/routes.rb
  get 'unsubscribe', to: 'welcome#unsubscribe'

# app/controllers/welome_controller.rb
  # GET /unsubscribe
  def unsubscribe
    referral_token_and_unsubscribe_group = Base64.decode64 params[:key] 
    referral_token = referral_token_and_unsubscribe_group[0..User::REFERRAL_TOKEN_LENGTH-1]
    @unsubscribe_group = referral_token_and_unsubscribe_group[User::REFERRAL_TOKEN_LENGTH..-1]
    if current_user
      if current_user.referral_token == referral_token
        @user = current_user
        @user.unsubscribe[ @unsubscribe_group] = "true"!
        redirect_to root_path, alert: "This key is for different user, please log out"
      if @user = User.find_by( referral_token: referral_token)
        # we found the user
        @user.unsubscribe[ @unsubscribe_group] = "true"!
        flash[:alert] = "Can't find user by this key. You need to signin in order to unsubscribe"
        redirect_to new_user_session_path and return

# db/migrate/_add_unsubscribe_to_user.rb
class AddUnsubscribeToUser < ActiveRecord::Migration
  def change
    add_column :users, :unsubscribe, :hstore, default: {}

Run rails in production mode

You probably need to set up database user for production env.

Also set config.force_ssl = false in config/environments/production.rb

Get template name

You can try directly in controller to byebug and try something like


Another is with patch ActionView::TemplateRenderer and this is only accessible in view.

# getting current template is not possible in rails
# patch taken from
# note 3th comment (Lukas) about exception notification issue
class ActionController::Base
  attr_accessor :active_template

  def active_template_virtual_path
    self.active_template.virtual_path if self.active_template

class ActionView::TemplateRenderer
  alias_method :_render_template_original_, :render_template

  def render_template(template, layout_name = nil, locals = {})
    if @view.controller && @view.controller.respond_to?('active_template=')
      @view.controller.active_template = template
      result = _render_template_original_( template, layout_name, locals)
      @view.controller.active_template = nil
      result = _render_template_original_( template, layout_name, locals)


Dealing with money with

Carrierwave for uploading

Store on server

cat >> Gemfile << HERE_DOC
gem 'carrierwave'
rails generate uploader Document
# we need just one field type string to store file url
rails g migration add_document_to_companies document:string
rake db:migrate
sed -i app/models/company.rb -e '/class Company/a \
  mount_uploader :document, DocumentUploader'
git add . && git commit -m "Adding carrierwave gem document uploader"

Replace f.text_field :document with f.file_field :document in your form. In view you can use company.document.url.

<%# app/views/companies/_form.html.erb %>
  <%  if @company.document.present?  %>
    <%= image_tag @company.document, class: 'image-small'%>
    <%= f.check_box :remove_document %>
  <% end %>
  <%= f.file_field :document %>
  <%# keep the file on reload in case of other validation errors %>
  <%= f.hidden_field :document_cache %>

# app/controllers/companies_controller.rb
  def company_params
    params.require(:company).permit(:document, :remove_document)

It is straightforward to use uploader in multiple fields. Also you can use single table field for multiple files (field type json) but than you need postgres database.

When you rendering json, than carrierwave will add nested url. Solution is render json manually with json.document_url company.document.url or to override uploader serilization with

# app/uploaders/document_uploader.rb
  def serializable_hash

Resizing is by adding mini magick and configure uploader. It works on Heroku too. You can process files, create new versions based on condition or process based on condition

echo "gem 'mini_magick'" >> Gemfile

# app/uploaders/document_uploader.rb
  include CarrierWave::MiniMagick
  process resize_to_limit: [200, 300]
  process resize_to_limit: [300, 300], if: :logo?
  def logo?(picture)
    # check if we mount_uploader :logo_url or something else
  version :thumb do
    process resize_to_fill: [200, 300]

Store on AWS S3

For Amazon S3 you need to set up your AWS keys in ~/.bashrc export AWS_ACCESS_KEY_ID=123123 and export AWS_SECRET_ACCESS_KEY=123123. Bucket should be created as standard USA bucket.

cat >> Gemfile << HERE_DOC
gem 'fog'
cat > config/initializers/carrierwave.rb << 'HERE_DOC'
CarrierWave.configure do |config|
  config.fog_credentials = {
    :provider               => 'AWS',
    :aws_access_key_id      => Rails.application.secrets.aws_access_key_id,
    :aws_secret_access_key  => Rails.application.secrets.aws_secret_access_key,
    :region                 => Rails.application.secrets.aws_region # us-east-1
  config.fog_directory  = Rails.application.secrets.aws_bucket_name

sed -i config/secrets.yml -e '/^test:/i \
  # aws s3\
  aws_bucket_name: <%= ENV["AWS_BUCKET_NAME"] %>\
  aws_access_key_id: <%= ENV["AWS_ACCESS_KEY_ID"] %>\
  aws_secret_access_key: <%= ENV["AWS_SECRET_ACCESS_KEY"] %>\
  # region is important for all non us-east-1 regions\
  aws_region: <%= ENV["AWS_REGION"] || "us-east-1" %>\

sed -i app/uploaders/document_uploader.rb -e '/storage :file/r \
  # storage :file\
  storage :fog'

git add . && git commit -m "Configure AWS S3"

Store directly on AWS S3 and upload the key to the server

You can put the direct_upload_form_for on any page, let’s use show:

cat >> Gemfile << HERE_DOC
# direct upload to S3
gem 'carrierwave_direct'

sed -i app/uploaders/document_uploader.rb -e '/DocumentUploader/a \
  include CarrierWaveDirect::Uploader'

sed -i app/uploaders/document_uploader.rb -e '/store_dir/c \
  # we do not use store_dir because of dirrect carrierwave\
  def store_dir_origin'

cat >> app/views/companies/show.html.erb << 'HERE_DOC'
<%= direct_upload_form_for @uploader do |f| %>
  <%= f.file_field :document %>
  <%= f.submit %>
<% end %>

sed -i app/controllers/companies_controller.rb -e '/def show/a \
   # @uploader = @company.document # do not use old since key will remain\
   @uploader =\
   # default key is /uploads/<unique_guid>/foo.png\
   # you can change, but use ONLY ONE folder ie "1/2/a.txt" -> "2/a.txt"\
   # it always adds prefix "uploads" so it does not need to be written\
   @uploader.key = "uploads/#{}-#{request.ip}/${filename}"\
   @uploader.success_action_redirect = company_url(@company)\
   if params[:key]\
     @company.document.key = params[:key]\!\
     # we need to reload since old key is there\
     @company = Company.find(\
     # or to redirect\
     redirect_to company_path(@company)\
   # you can call @company.remove_document! to remove from aws, but please\
   # reload after that with @company = Company.find('

sed -i config/initializers/carrierwave.rb -e '/^end/i \
  # max_file_size is not originally on carrierwave, but is added on CWDirect\
  # if file is greater than allowed than error is from Amazon EntityTooLarge\
  config.max_file_size = 20.megabytes  # defaults to 5.megabytes'

Carrier wave in seed

You can open file and use it in seed user.image url =, 'public/my_image.png')). But if your storage is fox than you can use user.remote_image_url_url = ''. Note that file will be downloaded and uploaded to your aws bucket so better is to set storage Rails.env.development? ? :file : :fog and use first method so it does not need to download file.


Wicked PDF

Generate pdf from html files. Just put in gemfile

cat >> Gemfile << HERE_DOC
# pdf generation from html
gem 'wicked_pdf'
gem 'wkhtmltopdf-binary'

And in your controller put the name of the downloaded file

class ThingsController < ApplicationController
  def show
    respond_to do |format|
      format.pdf do
        render pdf: "thing-#{params[:id]}"   # Excluding ".pdf" extension.

You can also set template to another file, layout to pdf and header

render template: 'periods/visits', pdf: "visits-#{@period.end_date}", layout: 'pdf', header: { right: '[page] of [topage]' }


# app/views/things/show.pdf.erb
<div class="alwaysbreak"></div>
<h1>Second page</h1>


<%# app/views/layouts/pdf.pdf.erb
<!doctype html>
    <meta charset='utf-8' />
    <%= stylesheet_link_tag wicked_pdf_asset_base64("pdf") %>
    <div id="content">
      <%= yield %>


// app/assets/stylesheets/pdf.scss
table, th, td {
  border: 1px solid black;
  border-collapse: collapse;
  padding: 5px;
  text-align: center;
* {
  font-size: 12px;
div.alwaysbreak { page-break-before: always; }

Since we did not include this style in asset pipeline, we need to precompile it:

# config/initializers/assets.rb
Rails.application.config.assets.precompile += ['pdf.css']

If you want to use existing html template than use param render pdf: 'name', template: 'show.html' and create app/views/layouts/pdf.html.erb and use helper classes to show hide content:

.pdf-hidden {
  display: none;
.html-hidden {
  display: none;


PDF reader

Style guide

toughtbot style guide

My style


If you need to search with association, for example User has many locations you can with joins to geocoded model on which near is defined.

near = Location.near('Paris, France')
users = User.joins(:locations).merge(near)

also if you need to get associated objects you can (location has many views)

near = Location.near([latitude, longitude], MAX_USER_DISTANCE_MILES)
  select("views.*"). # somehow we need this so we got views, instead of users
  includes(:sport). # so we can get without N+1
  where(sport: sport) # conditional on view

Localisation i18n translations

Tips To translate active record messages for specific attributes, you can overwrite messages for specific model and attributes (default ActiveRecord messages taken)

Also you can change format errors.format: Polje "%{attribute}" %{message} You can also see some default en translations.

And you can change attribute name имејл To translate also plurals you can use User.model_name.human(count: 2). For attributes you can use User.human_attribute_name("email") link

        zero: No dudes
        one: Dude
        other: Dudes

For form objects include ActiveModel::Model you should translate activemodel. For ApplicationRecord translate activerecord.

# config/locales/
        current_city: Који је твој град ?
        group_not_exists_for_age: Не постоји група (%{age}год) на овој локацији
              blank: Не може бити празно ?
        one: корисник
        other: корисници
        accusative: корисника
        some_customer_message: Моја порука

For custom errors can be different for each attribute or same. Can also accept param, for example

    errors.add :from_group_age, :group_not_exists_for_age, age: age For serbian you can provide pluralization

# config/locales/plurals.rb
serbian = {
  i18n: {
    plural: {
      keys: %i[one few many other],
      rule: lambda { |n|
        if n % 10 == 1 && n % 100 != 11
        elsif [2, 3, 4].include?(n % 10) && ![12, 13, 14].include?(n % 100)
        # elsif (n % 10).zero? || [5, 6, 7, 8, 9].include?(n % 10) || [11, 12, 13, 14].include?(n % 100)
        #   :many
        # there are no other integers, use :many if you need to differentiate
        # with floats
  sr: serbian,
  'sr-latin': serbian,
# config/initializers/pluralization.rb
require "i18n/backend/pluralization"
I18n::Backend::Simple.send(:include, I18n::Backend::Pluralization)
# config/locales/sr.yml
    # 1, 21, 31 ...
    one: %{count} порука је послата
    # 2, 3, 4, 22, 23, 24, 32, 33, 34 ...
    few: %{count} поруке су послате
    # all other integers: 5, 6, ... 9, 10, 11, 12, 13, 14 ... 20, 25, ...
    many: %{count} порука је послато

Note that you have to provide few translation for all words, since it could happend that count is 2 and translation is missing.

I18n.t 'sent_messages', count: 15

You can translate to any language with

I18n.t 'sent_messages', locale: :sr

Separate translations for models config/locales/sr.yml and views config/locales/views/sr.yml and include them with:

# config/application.rb
config.i18n.load_path += Dir[Rails.root.join('config', 'locales', '**', '*.{rb,yml}')]

Example for Serbian localizations translations:

# config/locales/sr.yml
    format: Поље "%{attribute}" %{message}
      blank: не сме бити празно
      invalid: није исправно
        required: мора постојати
        taken: је већ заузет
      user: корисник
        one: локација
        other: локације
        email: Имејл
        password: Лозинка
        password_confirmation: Потврда лозинке
        remember_me: Запамти ме

When you use .capitalize or .upcase than you need first to call .mb_chars. For example

=> "ž"

 => "Ž"

Some common words translations can be found

To translate with accusative you need to joins strings or use param in translation

module TranslateHelper
  # there are two ways of calling this helper:
  # t_crud 'are_you_sure_to_remove_item', item: @move
  # t_crud 'edit', User
  def t_crud(action, model_class)
    if model_class.class == Hash
      t(action, item: t("neo4j.models.#{model_class[:item].name.downcase}.accusative"))
      "#{t(action)} #{t("neo4j.models.#{}.accusative")}"

For model translation you can use human method

  <%= link_to User.size.to_s + ' ' + User.model_name.human(count: 2), admin_users_path %>

For non model you can use simple translation

  <%= link_to t('report', count: Message.), admin_reported_messages_path %>

Translate latin to cyrilic with You need to set language in config

# Gemfile
# translate cyrillic
gem 'cyrillizer'
# config/initializers/cyrillizer.rb
Cyrillizer.language = :serbian

In console

'my string'.to_cyr
 => "мy стринг"

Note that some chars looks the same but are not when rendered on html page

 # for example first line is not correct link a href
 <a href='%{confirmation_url}'>Поново пошаљи упутство за потврду</а>"
 <a href='%{confirmation_url}'>Поново пошаљи упутство за потврду</a>"


use cancancan and define all your actions in app/models/ability. If you want to use load_resource, you should separately define: index (with hash), show/edit/update/destroy (with block or hash), new/create (with hash). For nested resources just write parent association

Note that following next cannot rule will override a previous can rule, so it is enough to set can :manage, :all and than write what cannot :destroy, Project

Other autorization policy

Autoloading In development rails uses Constant Missing hooks to auto load new files from app/controllers helpers mailers and models If you need to load files from lib, you can config.autoload_paths += %W(#{config.root}/lib). Rails look for file name that is snake case of constant name, for example SeniorDeveloper should be defined in senior_developer.rb. You can require_dependency 'not_conventional_file_name'

There is ruby autoload :Jeep, 'Jeep' which is usefull since it will not require 'jeep' if Jeep is not used. If we use Jeep in a file, than it will required.

Action Cable example

Consumer is a client of a web socket connection that can subscribe to one or multiple channels. Each ActionCable server may handle multuple connections (it is created per browser tab).

rails g channel chat
# app/assets/javascripts/channels/
# app/channels/chat_channel.rb

You can broadcast with

ActionCable.server.broadcast("chat:#{}", data: data)
ChatChannel.broadcast_to chat, data: data

or in javascript = App.cable.subscriptions.create 'ChatChannel'{data: data})

You can call server method with @perform 'away', my_id: $('main').data('my-id')

You can not use devise current_user when rendering for actioncable, there will be an error

ActionView::Template::Error (Devise could not find the `Warden::Proxy` instance on your request environment.
Make sure that your application is loading Devise and Warden as expected and that the `Warden::Manager` middleware is present in your middleware stack.
If you are seeing this on one of your tests, ensure that your tests are either executing the Rails middleware stack or that your tests are using the `Devise::Test::ControllerHelpers` module to inject the `request.env['warden']` object for you.):


# app/helpers/page_helper.rb
module PageHelper
  def page_title(title)
    # this will add both page title and header below topnav
    content_for(:page_title) { title }
    content_for(:page_header) { title }

  def page_description(description)
    content_for(:page_description) { description.html_safe }

  def breadcrumb(list)
    @breadcrumb = list

  def get_breadcrumb_list
    @breadcrumb || []

# app/views/layouts/application.html.erb
  <ol class="breadcrumb">
    <% get_breadcrumb_list.each_with_index do |(text, link), i| %>
      <li class="<%= 'active' if i == get_breadcrumb_list.length - 1 %>">
        <% if link.present? %>
          <%= link_to link do %>
            <% if i == 0 %>
              <i class="fa fa-dashboard"></i>
            <% end %>
            <%= text %>
          <% end %>
        <% else %>
          <% if i == 0 %>
            <i class="fa fa-dashboard"></i>
          <% end %>
          <%= text %>
        <% end %>
    <% end %>

# app/views/customers/index.html
  page_title "Customers List"
  case params[:non_table_filter]
  when "registered_today"
    page_description "Registered Today"
  when "registered_this_month"
    page_description "Registered This Month"
  when "renewed_in_advance"
    page_description "Renewed In Advance"
  breadcrumb "Dashboard": dashboard_path, "Customers": nil

Maybe find can help

    :select => 'count(*) count, country', 
    :group => 'country', 
    :conditions => ['validated = ?', 't' ], 
    :order => 'count DESC',
    :limit => 5)
require File.expand_path('/home/orlovic/rails/myApp/config/environment', __FILE__)
puts User.all

def MyClass
  def self.issues
    @issues ||= Brand.find(1).issues.inject({}) do |hash, issue|
      hash[issue.title] = issue

If the value is falsy nil or false than you should not use ||= since it will have affect as =. You should check if instance variable is defined. You can also use begin end for multiline definition

class User < ActiveRecord::Base
def main_address
   return @main_address if defined? @main_address
    @main_address = begin
      main_address = home_address if prefers_home_address?
      main_address ||= work_address
      main_address ||= addresses.first # some semi-sensible default
  <%= f.check_box :renew, data: { url: customer_path(@customer), remote: true, method: :patch } %>
  <%= check_box_tag name, value, checked, data: { url: toggle_todo_path(todo), remote: true } %> for something that need test or we just want to showcase in one file onepage rails

zeus init
zeus start
touch config/boot
require 'net/http'
require 'uri'

def fetch(uri_str, limit = 10)
  # You should choose better exception.
  raise ArgumentError, 'HTTP redirect too deep' if limit == 0

  url = URI.parse(uri_str)
  req =, { 'User-Agent' => 'Mozilla/5.0 (etc...)' })
  response = Net::HTTP.start(, url.port) { |http| http.request(req) }
  case response
  when Net::HTTPSuccess     then response
  when Net::HTTPRedirection then fetch(response['location'], limit - 1)

print fetch('')

dhh tips for rails

module Account::Administered
  extend ActiveSupport::Concern

  included do
    has_many :administratorships, dependent: :delete_all do
      def grant(person)
        create_or_find person: person

      def revoke(person)

    has_many :administrators, through: :administratorships, source: :person

  def all_administrators
    administrators | all_owners

  def administrator_candidates
      where.not(id: administratorships.pluck(:person_id)). ownerships.pluck(:person_id)).

require ‘minitest/autorun’

class Test < Minitest::Test

def test_one_substring

s = ‘My name is John.’

assert_equal “My name is \e[John\e[0m.”, s.colorize(“John”, true)



def test_two_substrings

s = ‘John is my name, John.’

r = “\e[John\e[0m is my name, \e[John\e[0m.”

assert_equal r, s.colorize(“John”, true)



def test_no_found

s = ‘My name is John.’

assert_equal “My name is John.”, s.colorize(“Mike”, true)



def test_whole

s = ‘My name is John.’

assert_equal “\e[31mMy name is John.\e[0m”, s.colorize(s, true)



def test_return

s = ‘My name is John.’

assert_equal nil, s.colorize(‘John’)