I had to update a test that was very strict on the format. I realized that Capybara::Node::Simple doesn’t provide a selector for regex.

The page object doesn’t have a match? method

expect(page).to have_content('Name:Foobar') # ✅
expect(page).to match(/Name:Foobar/) # ❌

I needed to use the underlying object which isn’t body or content, this time it’s text, cool, cool, cool, coooool

expect(page.text).to match(/Name:\W+Foobar/) # ✅

The full diff:

[EDIT] A better solution for whitespaces normalize_ws

expect(page).to have_content('Name: Foobar', normalize_ws: true)

I still need to update the string to include a space but it feels a lot cleaner that way. Thanks to my coworker for the hint.

There is the self explanatory one: add versioning to the schema file.

And the noisy change of every single datatime field's precision attribute.

Looking at this, I find out that the precision default to 6. It explains why some attributes got precision: nil and some other got their precision: 6 removed.

https://github.com/rails/rails/blob/aed8feae3b7a3f7df59de69355cc3bda1d5479d6/activerecord/lib/active_record/connection_adapters/abstract/schema_statements.rb#L1322

This was a change made in 2019 and released from Rails 7.0 in December 2021. So, 2.5 years after but I'm happy to discover that and sharing it.

https://github.com/rails/rails/commit/57015cdfa2083351f64a82f7566965172a41efcb#diff-868f1dccfcbed26a288bf9f3fd8a39c863a4413ab0075e12b6805d9798f556d1R714-R716

For people following here, Rails 7.1 is already planned and is already passing the entire test suite. Hopefully, it won't take long.

Cheers!

Doc: https://docs.appsignal.com/logging/platforms/cloudwatch.html

Same here! But you are in luck, here's the script I used to do all steps at once!

You only need the AppSignal token (given after you created a new CloudWatch source) and the CloudWatch group log name (not the ARN).

The CloudWatch template: https://gist.github.com/D-system/6dc155100de058b8a7c4c58b136d5bf1

I provided their support with the template and invited them to add the CloudFormation template to their doc 😀 Let's see if it'll be added.

My setup:

• 1 main database

• 1 replica

What I did not understand:

• Where to put the connects_to setting

• Proper/clean configuration in database.yml

The database.yml

Notes:

• All environment needs primary and primary_replica otherwise Rails will raise an exception (at boot).

• The *_replica needs replica: true

• I did not put primary and primary_replica inside default because any change in a sub-object will erase all properties (I think)

• In that example, only production has a proper replica with different settings

default: &default
  adapter: trilogy
  encoding: utf8mb4
  pool: 5
  database: <%= ENV.fetch('DB_NAME') { 'db_name' } %>
  username: <%= ENV.fetch('DB_USER') { 'user' } %>
  password: <%= ENV.fetch('DB_PASS') { 'password' } %>
  host: <%= ENV.fetch('DB_HOST') { 'db.example.com' } %>
  port: <%= ENV.fetch('DB_PORT') { '3306' } %>

development:
  primary:
    <<: *default
  primary_replica:
    <<: *default
    replica: true

test:
  primary:
    <<: *default
  primary_replica:
    <<: *default
    replica: true

staging:
  primary:
    <<: *default
  primary_replica:
    <<: *default
    replica: true

production:
  primary:
    <<: *default
  primary_replica:
    <<: *default
    username: <%= ENV.fetch('DB_USER_REPLICA') { 'user_replica' } %>
    password: <%= ENV.fetch('DB_PASS_REPLICA') { 'password_replica' } %>
    host: <%= ENV.fetch('DB_HOST_REPLICA') { 'db-replica.example.com' } %>
    replica: true

Model set-up: connects_to

I thought I could use connects_to directly on a model (eg: User) but no, it should be on an abstract_class or ActiveRecord::Base (what the exception says) but I didn't understand it. I thought I should put abstract_class on the User but no, it should the parent class ApplicationRecord or anything other parent class (eg: AdminApplicationRecord for admin user).

class ApplicationRecord < ActiveRecord::Base
  self.abstract_class = true
  connects_to database: {
    writing: :primary,
    reading: :primary_replica
  }
end

If it helped, please share it.

Have fun, bye.

./bin/rails routes --unused

For Rails 6.0, 6.1 and 7.0, below is a script I used in my project to find out unused routes. Apparently, I barrow it from ShakaCode. Thanks to them.

In conjunction with the self contained rails script shown before, here's the full modified script:

#!/usr/bin/env ruby
APP_PATH = File.expand_path("../../config/application",  __FILE__)
require File.expand_path("../../config/boot",  __FILE__)
require APP_PATH

Rails.application.require_environment!

unused_routes = {}

Rails.application.routes.routes.each do |r|
  name = r.requirements[:controller].to_s.camelize
  action = r.requirements[:action].to_s
  controller = "#{name}Controller"

  if Object.const_defined?(controller) && !controller.constantize.new.respond_to?(action)
    unless Dir.glob(Rails.root.join("app", "views", name.downcase, "#{action}.*")).any?
      unused_routes[controller] = [] if unused_routes[controller].nil?
      unused_routes[controller] << action
    end
  end
end

pp unused_routes

The script is in the file bin/unused_routes. And it's an executable (chmod +x bin/unused_routes).

With that, you only need to run the command below. No more need to remember or search for it, everything lives in your repository.

./bin/unused_routes

Magical!

# inline script
./bin/rails runner "puts Rails.env"

# file
./bin/rails runner path/to/file.rb

If you want to have a self container, here's what you need (eg: filename bin/puts_env):

#!/usr/bin/env ruby
APP_PATH = File.expand_path("../../config/application",  __FILE__)
require File.expand_path("../../config/boot",  __FILE__)
require APP_PATH

Rails.application.require_environment!


###### Your script below
puts Rails.env

Don't forget to make the file executable:

chmod +x bin/puts_env

Now you can run the script, no more need to remember how to run it, just do:

./bin/puts_env

Magical!

Input

abc,abc
abc abc
abc

I just want to split on , , a space ... and a new line. Great

I won't show you an horrible .split with a .map then redo a .split again.

Use regexp

regexp = /[ ,\n]/
input.split(regexp)

It is annoying to read (even that simple) and it did not work for what I wanted. Remember the "new line", as a html input/textarea is not an \n but an \r\n .

You can change the regexp to /[ ,\r\n]/ but it split twice, once for \r then once more for \n. And the readability is lower.

Solution: Union!

regexp = Regexp.union([" ", "\r\n", "\n", ","])
input.split(regexp)

You give the string you want the regexp to split on and it generates the proper Regexp for you.

It won't solve all but for me, it solve most of my simple cases with code that I use everyday: a dead simple list of string.

Share it!

Cheers, Thomas

https://github.com/heartcombo/devise

For some sensitive application, you want to logout the user aggressively.

For example: a banking app. You want users to re-login if inactive for more then 30 minutes but keep them logged in if active.

Add :rememberable and :timeoutable to the devise method on the model.

# app/models/user.rb

class User < ActiveRecord::Base
  devise :database_authenticatable,
         :rememberable,
         :timeoutable
end

The main configuration to change are: remember_for and timeout_in.

# config/initializers/devise.rb

Devise.setup do |config|
  # [...]

  # ==> Configuration for :rememberable
  # The time the user will be remembered without asking for credentials again.
  config.remember_for = 1.day

  # Invalidates all the remember me tokens when the user signs out.
  config.expire_all_remember_me_on_sign_out = true

  # ==> Configuration for :timeoutable
  # The time you want to timeout the user session without activity. After this
  # time the user will be asked for credentials again. Default is 30 minutes.
  config.timeout_in = 30.minutes

  # [...]
end

That's all 🎉

Please share. If you have any suggestions, please comment.

# .irbrc
module ColorString
  refine String do
    def cyan = "\e[36m#{self}\e[0m"
    def bg_red = "\e[41m#{self}\e[0m"
    def bg_yellow = "\e[43m#{self}\e[0m"
    def bg_green = "\e[42m#{self}\e[0m"
  end
end

if defined?(Rails)
  using ColorString

  if Rails.env.production? || Rails.env.staging?
    IRB.conf[:SAVE_HISTORY] = false
  end

  project_name = File.basename(Dir.pwd).cyan
  environment = case Rails.env
                when "development"
                  Rails.env.bg_green
                when "staging"
                  Rails.env.bg_yellow
                when "production"
                  Rails.env.bg_red
                end
  prompt = "#{project_name} [#{environment}]"

  IRB.conf[:PROMPT] ||= {}
  IRB.conf[:PROMPT][:RAILS] = {
    PROMPT_I: "#{prompt} %03n > ",
    PROMPT_S: "#{prompt} %03n * ",
    PROMPT_C: "#{prompt} %03n ? ",
    RETURN: "=> %s\n"
  }

  IRB.conf[:PROMPT_MODE] = :RAILS
end

begin
  require "awesome_print"
  AwesomePrint.irb!
  rescue
end

It's possible with stickyScroll. The default contrast is low but you can change them in colorCustomizations.

// settings.json
    "editor.stickyScroll.enabled": true,
    "workbench.colorCustomizations": {
        "editorStickyScroll.background": "#5d5252",
        "scrollbar.shadow": "#fcfcfc"
    }

Issue: I have an old gem that reference an relative image in the CSS background-image: url("icon.png")) which once compile it become:

• Sprockets 3: background-image: url("icon.png")) (it's a relative URL)

• Sprockets 4: background-image: url("/icon.png")) (the / means root URL, eg http://example.org/icon.png)

• With the fix below: background-image: url("./icon.png")) (./ is also a relative URL)

Notes about relative URL:

• If the CSS file is http://example.com/application.css

  • then the icon.png is also at the root: http://example.com/icon.png

• If the CSS file is in a gem (for me it was CKeditor), the CSS file is http://example.com/assets/ckeditor/skins/moono-lisa/editor.css

  • then the icon.png is resolved as http://example.com/assets/ckeditor/skins/moono-lisa/icon.png

# config/initializers/sprockets.rb

# Issue exists: https://github.com/rails/sprockets-rails/issues/507
# Fix from PR: https://github.com/rails/sprockets-rails/pull/511
# It will not get merge as Sprockets is phased out.
module Sprockets
  module Rails
    # Resolve assets referenced in CSS `url()` calls and replace them with the digested paths
    class AssetUrlProcessor
      REGEX = /url\(\s*["']?(?!(?:\#|data|http))(?<relativeToCurrentDir>\.\/)?(?<path>[^"'\s)]+)\s*["']?\)/
      def self.call(input)
        context = input[:environment].context_class.new(input)
        data    = input[:data].gsub(REGEX) do |_match|
          path = Regexp.last_match[:path]
          # Prevent non-digest relative path from being converted to root absolute path
          orig_path = path
          path = context.asset_path(path)
          path = (path == "/#{orig_path}") ? ".#{path}" : path
          "url(#{path})"

        end

        context.metadata.merge(data: data)
      end
    end
  end
end

Caveat: the original file did not change therefore the digest did not either, until the browser invalid the the cache and download the new CSS, users will have the old CSS cached by the browser

Please share and comment on how I can write it better.

# .github/workflows/my_very_special_job.yml
jobs:
  my_very_special_job:
    steps:
      - name: Setup Ruby
        uses: ruby/setup-ruby@v1
        with:
          bundler-cache: true    ###### That's the line