Home
Check out aiproxy.pro


Rails 7 Notes

Perform the same JS on document ready and turbo frame load

function myFn() { ... }  
document.addEventListener("turbo:frame-load", myFn);  
document.addEventListener("DOMContentLoaded", myFn);

Get signups by day

User.group("DATE(created_at)").count

How to run scripts

Use rails runner. Place scripts in the scripts folder, then:

./bin/rails runner script/backfill.rb --environment=production

Rails check if ActiveRecord model is yet to be saved to DB

my_model.new_record?

Rails URL helpers with param other than id

It’s not enough to specify the param in config/routes.rb, such as:

resources :projects, param: :my_uuid

I also have to set

class Project < ApplicationRecord  
  def to_param  
    my_uuid  
  end  
end

Rails has_one gotcha

If model A has_one B, and you use:

b = a.build_b()

Any existing a.b is quietly deleted, and this occurs even if the new b does not pass validation.

Rails URL helper weirdness

Use variadic call instead of passing an array. Note the rogue %2F:

(ruby) project_device_check_configuration_path([@project, @device_check_configuration])  
# => "/projects/abdbf565%2F/device_check_configuration"

versus

(ruby) project_device_check_configuration_path(@project, @device_check_configuration)  
# => "/projects/abdbf565/device_check_configuration"

How to inspect data in the test db

Rails uses transactions by default in test cases. If I breakpoint inside a test and inspect the the contents of the
test DB in a mysql client, I find no records. To temporarily disable transactions so that I can inspect DB contents, use:

class MyTest < ActiveSupport::TestCase  
  self.use_transactional_tests = false  
  # ...  
end

How to create a model that is not backed by the DB

class ProjectUIModel
include ActiveModel::API

attr_accessor :name  
validates_presence_of :name

end

I wanted this to use the rails (ActionPack) form helpers with a UI model.
See the full API description at ~/.rvm/gems/ruby-3.2.2/gems/activemodel-7.1.3.4/README.rdoc

How to eagerly load associations

Model.includes(:my_association).first

Nested:

Model.includes([first_level: [second_level: :third_level]])

How to build associated records

For has_many and variants, use

my_model.my_associations.build(...)  
# or  
my_model.my_associations.create(...)

For has_one use

my_model.build_my_association(...)  
# or  
my_model.create_my_association(...)

Update ActiveRecord fields

Run validations and callbacks:

my_model.update(some_field: 1)

Skip validations and callbacks:

my_model.update_attribute(:some_field, 1)

Dump a rails model to json with assocations

my_model.to_json(include: { my_association: {}})

or

ActiveSupport::JSON.encode(my_model, include: { my_association: {}})

How to run a specific migration programmatically

migration_version = 20240603120712 # Replace with your specific migration version  
migration_context = ActiveRecord::MigrationContext.new('db/migrate', ActiveRecord::SchemaMigration)  
migration_context.run(:down, migration_version)  
migration_context.run(:up, migration_version)

Related: to quiet the logs when running migrations programmatically, especially if inserting a bunch of seed data, use:

ActiveRecord::Schema.verbose = false

Source: https://stackoverflow.com/a/69011649/143447

How to run git bisect on a rails regression

https://github.com/rails/rails/issues/51910#issuecomment-2130296578

Skip tests and builder in generator

-skip-tests --skip-jbuilder

Custom validations

validate do  
    errors.add(:some_field, 'Error message') unless my_condition  
end

Or

validate :something_custom  

private  
def something_custom  
  if my_condition  
    errors.add(:some_field, "Error message")  
  end  
end

Rails Turbo Frame gotchas

  1. In the action that I’m serving the partial page contents from, remember to skip layout rendering:

    def my_action
    # Define my_action.html.erb and populate it with partial contents
    render :layout => false
    end

  2. In the UI element that kicks off the fetch of partial content, specify the turbo frame to update:

    <%= turbo_frame_tag “frame-to-update” do %>
    Placeholder
    <% end %>

    <%= link_to ‘Update partial contents’, my_action_path, data: {turbo_method: :get, turbo_frame: “frame-to-update” } %>

Or, without rails form helpers:

<turbo-frame id="frame-to-update">  
    Placeholder  
</turbo-frame>  

<a data-turbo-method="get" data-turbo-frame="frame-to-update" href="<%= my_action_path %>">Update partial contents</a>

See all rake tasks

rails --tasks

Update rails

Get state of model before last save

Use the after_save hook. Say the model has an attribute called name,
inside the hook use name_before_last_save to get the previous value.

Tail puma log in production (AL2023)

journalctl -fu puma

Do not db:rollback in production

I have accidentally rolled back farther than intended, and then I have a production data fill on my hands.
Instead, make sure I use STEP=1. Or better yet be explicit with:

rake db:migrate:down VERSION=20240303...

Then, to go the other direction:

rake db:migrate:up VERSION=20240303...

Or:

rake db:migrate:redo VERSION=20240303...

How to see the status of production migrations

RAILS_ENV=production ./bin/rails db:migrate:status

Where is migration status stored in mysql?

The schema_migrations table:

MariaDB [aiproxy_production]> select * from schema_migrations;

How to run a migration in code

Careful with this. It will not modify the schema_migrations table:

./bin/rails c --environment=production  
require "./db/migrate/20240303150054_create_dynamo_tokens_table.rb"  
CreateDynamoTokensTable.new.down

Where is the validation API defined?

activemodel-7.0.7/lib/active_model/validations/validates.rb line 106

How to view active record errors

my_model.errors.full_messages

How to force quit puma in dev

Occassionally I find myself needing to do this:

rm aiproxy/dashboard/tmp/pids/server.pid  
pkill puma

If that doesn’t work, try:

lsof -i :3000  
:: find the ruby process  
kill -9 <id>

Or

ps aux | grep puma | grep -v grep | awk '{print $2}' | xargs kill

How to clean up default rails routes

Follow these instructions to get rid of all the /rails/conductor/action_mailbox routes:
https://www.youtube.com/watch?v=IDsYWrsmO9g

How to clear cache

Run rails tmp:clear and remove everything in public/assets

(hack to include routes in activerecord model)

class Router  
  include Rails.application.routes.url_helpers  

  def self.default_url_options  
    ActionMailer::Base.default_url_options  
  end  
end  

Then use `Router.new.my_model_url(@my_model)`

Source: https://stackoverflow.com/a/54542949/143447

(rails destroy confirmations)

Button:

<%= button_to "Delete", @my_model, form: { data: { turbo_confirm: "Are you sure?" } }, method: :delete %>

Link:

<%= link_to "Delete", my_path, data: {turbo_method: :delete, turbo_confirm: 'Are you sure?'} %>

(render text)

render :text is no longer a thing. Use render plain: "hi" instead

(testing locally on mobile)

Start the server with ./bin/rails s -b 0.0.0.0
Use the following to get my 192.168 IP:

ifconfig | grep 'inet '

Browse to 192.168.X.Y:3000

(turbo, link to helpers, delete method, post method, deprecated)

This is no longer possible in Rails 7:

link_to "Remove something", my_path, method: :delete

Instead, turbo must be used:

link_to "Remove something", my_path, data: { turbo_method: :delete }

Or, use a form instead:

<%= button_to "Destroy", @my_model, method: :delete %>

Or, write a data-turbo-method attribute on the html with the desired verb (post in this case):

<a data-turbo-method="post" href="/users?color=red">enter as red</a>

(debugging action cable javascript)

cd <my-proj>  
vim "$(bundle show actioncable)/app/assets/javascripts/actioncable.esm.js"  
:: restart rails server after making modifications!

(rails 7, test redis connection in production)

./bin/rails c --environment=production  
irb> r = Redis.new  
irb> r.ping  
=> "PONG"

(rails 7, generator, add index, add column)

Add an indexed column to the db with one generator:

rails g migration add_index_to_rooms uuid:string:index

(rails 7, debug web socket, firefox)

View the ActionCable activity at Dev tools > Network > Tap on cable > WS tab > Response subtab

(rails 7, debug importmap)

View importmap with:

./bin/importmap json

Keep playing with syntax in config/importmap.rb until importmap json shows the importmaps I expect.

Pin an external dependency with:

./bin/importmap pin <dep-name>

(rails 7, add jquery)

Add to importmap:

./bin/importmap pin jquery

Modify application.js to include:

import jquery from "jquery"  
window.$ = jquery  
window.jQuery = jquery

(rails 7, styles, scaffolding, dhh demo)

https://www.youtube.com/watch?v=mpWFrUwAN88

Highlights:

(rails 7, actioncable, websocket, per user)

Set up the cable so that it streams to the current user:

// app/channels/application_cable/connection.rb  
// Source: https://guides.rubyonrails.org/action_cable_overview.html  
module ApplicationCable  
  class Connection < ActionCable::Connection::Base  
    include ActionController::Cookies  
    identified_by :current_user  

    def connect  
      self.current_user = find_verified_user  
    end  

    def disconnect  
      puts "#{self.class} Disconnected"  
    end  

    private  
      def find_verified_user  
        if verified_user = User.find_by_id(cookies.encrypted['_YOUR_APP_NAME_session']['user_id'])  
          verified_user  
        else  
          reject_unauthorized_connection  
        end  
      end  
  end  
end  


// my_channel.rb  
class MyChannel < ApplicationCable::Channel  
    def subscribed  
      stream_for current_user  
    end  
    ...  
 end

Send notifications to specific user:

// https://stackoverflow.com/a/43943322/143447  
MyChannel.broadcast_to(user, { notification: 'Test message' })

Get all cables for a specific user:

ActionCable.server.remote_connections.where(current_user: current_user)

Question to experiment with:
Is the web socket torn down when navigating between pages?

URL helpers are not available by default in ActionCable.
Use this:

class MyChannel < ApplicationCable::Channel  
  include Rails.application.routes.url_helpers

(rails, actioncable)

Generator to create actions in a channel:

rails g channel <channel-name> <action-name>

This modifies the import map at config/importmap.rb to add the actioncable dependency as an esm module,
and modifies app/javascript/application to import everything under the app/javascript/channels dir.

I can get a reference to a connected cable by modifying <channel-name>_channel.js to contain:

window.myChannel = consumer.subscriptions.create("MyChannel", { ...

Restart the rails server, refresh firefox, open dev tools (cmd+opt+i), tap network, tap WS.
There will be a connected web socket.
Open the console, I can send <action-name> with:

myChannel.<action-name>()

For example, following the DHH tutorial for action cable, if I use the generator rails g channel room speak, I can:

myChannel.speak()

Add an argument by modifying the speak definition in room_channel.js. E.g.

// room_channel.js  
speak: function(message) {  
  return this.perform('speak', message);  
}

then on ruby side:

// room_channel.rb  
def subscribed  
  stream_from "room_channel"  
end  

def speak(cable_data)  
  puts "Room channel received #{cable_data['message']}; echoing..."  
  ActionCable.server.broadcast 'room_channel', {message: cable_data['message']}  
end

and back on client:

// room_channel.js  
received(data) {  
  alert(data['message'])  
},

finally, punching myChannel.speak({message: "hello world"}) into the dev tools console should pop an alert message.

There is also a bit of this vid (16:40) where he shows that UI can be updated from an async job using ActionCable. (background job, async job, actioncable, action cable)

Source for Rails 5 ActionCable demo (DHH): https://www.youtube.com/watch?v=n0WUjGkDFS0
My stackoverflow question: https://stackoverflow.com/questions/76964623/what-is-the-equivalent-of-app-cable-for-rails-7
See ~/dev/actioncable_experiment

(rails 7, hotwire)

Clip from hotwire demo:
“Turbo streams deliver page changes over web sockets or in response to form
submissions using just html and a set of crud-like action tags. The tags let
you append, prepend, or replace and remove any target dom element from the
existing page. They’re strictly limited to DOM changes though, no direct
javascript invocations. If you need more than DOM change, connect a stimulus
controller.” minute 5 here: https://www.youtube.com/watch?v=eKY-QES1XQQ

(rails, stimulus)

You can think of it this way: just like the class attribute is a bridge connecting HTML to CSS, Stimulus’s data-controller attribute is a bridge connecting HTML to JavaScript.
https://stimulus.hotwired.dev/handbook/introduction

(rails credentials, gotcha, caution)

Do not try to set an environment variable for environment! This:

RAILS_ENV=development EDITOR='vim' bin/rails credentials:edit

is not the same as this:

EDITOR='vim' bin/rails credentials:edit --environment=development

(rails credentials, store credentials)

Edit credentials with:

rails credentials:edit --environment=development

Environment credentials are stored at:

config/credentials/<environment>.yml.enc

Environment credentials key is stored at (do not check this in!):

config/credentials/<environment>.key

If I edit development credentials:

./bin/rails credentials:edit --environment=development  

stripe:  
  publishable_key: <snip>  
  secret_key: <snip>

I can reference the secret at runtime with:

Rails.application.credentials.stripe.secret_key

The master key is stored at (do not check this in!):

config/master.key

Change master key: https://stackoverflow.com/a/59993704/143447

In production, running this:

EDITOR="/usr/bin/vim" bin/rails credentials:edit

changes the file contents at config/credentials.yml.enc

I prefer not to use credentials:edit without an environment,
instead relying upon config/credentials/development.* and config credentials/production.*

(rails credentials, debugging)

If in production I get the error:

ArgumentError: Missing `secret_key_base` for 'production' environment, set this string with `bin/rails credentials:edit`

Either create an env variable called SECRET_KEY_BASE or add secret_key_base to production.yml.enc with:

./bin/rails credentials:edit --environment=production

(rails credentials in console)

Rails.application.credentials.config

(rails credentials, how does this work, rake secret)

I don’t understand how this works.
Why does ./bin/rake secret dump a different secret to the console each time?
Oh, it is not dumping secret_key_base, it is just spitting out a random string for use as a secret.
See the task description with rake -T secret

Read ./bin/rails credentials:help
In development and testing, Rails.application.secrets.secret_key_base is derived from the app name.
In production, I add secret_key_base to production.yml.enc

(rails, error, localhost, add to hosts, https)

To serve at https://localhost without host blocking,
add config.hosts.clear to config/environments/development.rb