rails_new
rails_new copied to clipboard
A thoughtfully designed template for building modern Rails apps. Get started in minutes instead of hours 🔥🚀
rails_new
A thoughtfully designed template for building modern Rails apps 🔥
Get started on your new app within minutes instead of hours 🚗💨
- rails_new
- Getting started
- Optional configuration options
- Environment variables
- Contents
- Base system
- General
- Frontend
- Development
- Test
- Production
- Removed
- Notes
AccountvsUser- Heroku Buildpacks
- React
- TypeScript
- Hot Module Replacement (HMR)
- JavaScript/TypeScript Tests
- Polyfills
- Custom classes
Gemfile.dev/Gemfile.dev.lockRack::RejectTracemiddleware- Favicons
- Docker
- Usage examples:
- TODO
- License
Getting started
To get going clone this repository and perform the following steps:
- Run
rails credentials:editto re-generateconfig/master.keyand createconfig/credentials.yml.enc. - If your application requires
ActiveStorage, runrails active_storage:installto generate a migration that creates the necessary tables. Userails db:migrateto run the migration. - You can now run
bin/configure, which will help you in configuring the template.
As an alternative to running the script you can perform all of the following steps manually.
- Change application name in
config/application.rb. - Update
database.ymlto reflect the new application name. - Update
TODOitems inconfig/environments/production.rb. - If you plan on using Figaro, copy
config/application.yml.exampletoconfig/application.yml. - ESLint is preconfigured for modern JS with React support (using Prettier). If you want to use it install packages with
npm install/yarn [install], otherwise remove.eslintrcandpackage.json. - If you don't plan on tracking the template, you can remove the remote, otherwise rename it and add your new remote as appropriate.:
$ git remote rename origin rails_new $ git remote add origin ...
Optional configuration options
- If you want to use Sentry, you'll need to:
- Create two projects under your organization on Sentry. This way you can keep frontend and backend errors separate.
- Set
SENTRY_DSN_BACKENDandSENTRY_DSN_FRONTENDenvironment variables. Both are optional - errors will only be reported if the respective variable is set.
- New Relic is pre configured in
config/newrelic.yml, but you need to comment in the environment variables for it work on Heroku (lines 10 and 17). - The app is preconfigured for Google Analytics, just add
GOOGLE_ANALYTICS_IDto the environment. - We prefer to use vanilla Sidekiq for worker/queue management. If you prefer to use ActiveJob please see the configuration/options that were removed in 59cf38d.
Environment variables
| Variable | Comment |
|---|---|
| SENTRY_DSN_BACKEND | Used to report backend errors to Sentry. |
| SENTRY_DSN_FRONTEND | Used to report frontend errors to Sentry. |
| BLOCK_HTTP_TRACE | Disables HTTP TRACE method if set to true/t/1 |
| BUNDLE_GEMFILE | Useful when using a Gemfile.dev |
| DATABASE_URL | Used for production env, automatically set by Heroku |
| GOOGLE_ANALYTICS_ID | Will be added to the main application layout if set |
| HOST | Your base URI, e.g. https://myapp.herokuapp.com |
| NEW_RELIC_APP_NAME | Used in config/newrelic.yml |
| NEW_RELIC_LICENSE_KEY | Used in config/newrelic.yml |
| PORT | Port Puma will listen on, defaults to 3000 |
| RACK_TIMEOUT_SERVICE_TIMEOUT | Limit for Rack::Timeout, defaults to 15 seconds |
| RAILS_LOG_TO_STDOUT | Set by Heroku Ruby buildpack, set manually on other platforms if needed |
| RAILS_MAX_THREADS | Number of Puma threads, defaults to 5 |
| REDIS_URL | Used in config/cable.yml |
| WEB_CONCURRENCY | Number of Puma workers. We default to threads only, no workers |
Contents
All of the following have been installed and pre-configured:
Base system
- Ruby
- Ruby on Rails
- PostgreSQL >= 9.2
- pg for
ActiveRecord - NodeJS
NOTE: We recommend that you install and manage these system dependencies using a combination of Homebrew (Postgres), asdf-vm (Ruby, NodeJS) and Bundler (Ruby on Rails). If you are starting fresh on a new system you may want to manage these package managers with a script like the thoughtbot laptop script with the option to add your own opinionated extensions, e.g. this. Your code base should be under version control, why not your system toolkit and configuration as well?
General
Frontend
All of these are managed by yarn.
Development
- Brakeman
- bullet
- letter_opener
- memory_profiler
- newrelic_rpm
- nullalign
- pry
- pry-byebug
- pry-doc
- rack-mini-profiler
- RuboCop
Test
Rspec has been preconfigured for Rails 5.1+ system tests.
- No need to
require 'rails_helper, we do it in.rspec - bundler-audit (runs on CircleCI)
- capybara
- factory_bot_rails
- shoulda-matchers
- webmock
Production
- sentry-ruby
- heroku-deflater
- lograge
- rails_12factor
- rack-timeout: Configured via env variables, see documentation
- secureheaders
Removed
The following default Rails gems have been removed:
Notes
Account vs User
Authentication concerns (your typical Devise configuration) are handled by the Account model. To connect this to one of several potential user roles the polymorphic authenticatable relationship is used.
Heroku Buildpacks
Note that for everything to work properly on Heroku you need to set up your buildpacks like this:
heroku buildpacks:clear
heroku buildpacks:add heroku/metrics
heroku buildpacks:set heroku/nodejs
heroku buildpacks:add heroku/ruby
React
React support has been preconfigured together with react-rails.
When creating a new component you want to mount in the view, place it under the views folder. react-rails will perform module lookup relative to that folder and automatically require it under the hood so it gets included in the application pack. For example, calling the following helper in your view file:
react_component("dashboard/properties_list")
will require views/dashboard/properties_list and mount it in place of the helper element. See the docs for details about working with react-rails.
TypeScript
TypeScript is supported out of the box.
It is important to note that TypeScript code is loaded by babel-loader. Because of that, you need to perform type checking in a separate process. This could either be your editor, or running yarn types-watch in your terminal to compile your code in watch mode.
This will not interfere with HMR (see below) as TypeScript compiler is instructed only to type-check your code, not to emit compiled modules.
For extra safety, there is a CI build step that runs tsc to make sure TypeScript code compiles
Hot Module Replacement (HMR)
HMR is also supported out of the box. All you need to do is hot-export your module, like so:
import React from "react";
import { hot } from "react-hot-loader/root";
class Welcome extends React.Component {
render() {
return <h1>Hello, {this.props.name}</h1>;
}
}
export default hot(Welcome);
This will allow you to continue working on your JS code without losing application state in the browser.
JavaScript/TypeScript Tests
The repo comes pre-configured with jest. You can write your tests in JavaScript or TypeScript.
We're following the convention of placing tests next to the file they're testing. They will be run as long as they have test, (e.g. MyComponent.test.ts) in the name.
You are encouraged to place your testing utility files under app/javascript/test.
To run your tests, run:
yarn test
Polyfills
Polyfills are included automatically thanks to @babel/preset-env. See babel.config.js for configuration options.
Custom classes
ApplicationDecorator: lightweight alternative to Draper or similar gems.ApplicationForm: Minimal form class based onActiveModel.ApplicationPresenter: a subclass ofApplicationDecoratorfor presenters, includes tag helpers.
All custom classes are fully documented with yard and come with generators.
Use yard doc to generate documentation and yard server --reload or yard server --gems to start a local documentation server.
Gemfile.dev / Gemfile.dev.lock
If you want to add specific gems for development that may not be interesting for other developers, you can add a Gemfile.dev (ignored by our .gitignore). Gems listed there can be installed with bundle install --gemfile Gemfile.dev and the resulting lock file is gitignored too.
Example Gemfile.dev:
source 'https://rubygems.org'
eval_gemfile 'Gemfile'
gem 'awesome_print'
The eval_gemfile line will ensure that all gems from your regular Gemfile will be included too. The BUNDLE_GEMFILE variable can be used to let Bundler now which gemfile to use:
BUNDLE_GEMFILE=Gemfile.dev rails c
Rack::RejectTrace middleware
There's a custom middleware (Rack::RejectTrace) for completely disabling the HTTP TRACE method as required by certain security audits. It can be enabled via the BLOCK_HTTP_TRACE environment variable.
Favicons
Favicons were generated with Real Favicon Generator, consider using the same tool when replacing them for your project.
Docker
For those wishing to use Docker for development the whole app has been dockerized and the setup is fairly well-documented. Features:
- Images use Alpine Linux to keep their size small.
docker-compose.ymlsets up and starts Postgres, Redis, Sidekiq, Rails and the Webpack dev server.- Uses a persistent bundle cache, so there's no need to rebuild the image to add gems.
- Persistent volumes for Postgres and Redis.
- No exposed ports except for Rails (mapped to port
3000by default)
The following files relate to our Docker setup:
Dockerfile: main setup for theappcontainerdocker-compose.yml: Ties Postgres, Redis andapptogetherdocker-entrypoint.sh: Custom entry point to facilitate bundle cachingProcfile.docker: Used by the entrypoint script to bring up services.dockerignore: similar to.gitignore, specifies files we don't want copied into the container
Usage examples:
Start the environment and build the images if necessary:
$ docker-compose up --build
Building app
[Step 1/12 : FROM ruby:2-alpine
---> 8302cc790fbf
Step 2/12 : RUN apk update && apk add --update --no-cache build-base chromium chromium-chromedriver git imagemagick libxml2-dev libxslt-dev nodejs tzdata postgresql-dev
---> Using cache
---> 9d1d0b398c26
Step 3/12 : RUN bundle config build.nokogiri --use-system-libraries
---> Using cache
---> 0a5ca06d7700
Step 4/12 : WORKDIR /app
---> Using cache
---> c61498ba7e64
[...]
Start the environment without (re-)building images (add -d to daemonize):
$ docker-compose up
Starting rails_new_postgres_1 ... done
Starting rails_new_redis_1 ... done
Starting rails_new_app_1 ... done
Attaching to rails_new_postgres_1, rails_new_redis_1, rails_new_app_1
[...]
Stop the environment but keep the containers:
$ docker-compose stop
Stopping rails_new_app_1 ... done
Stopping rails_new_postgres_1 ... done
Stopping rails_new_redis_1 ... done
Stop the environment and remove the containers:
$ docker-compose down
Stopping rails_new_app_1 ... done
Stopping rails_new_postgres_1 ... done
Stopping rails_new_redis_1 ... done
Removing rails_new_app_1 ... done
Removing rails_new_postgres_1 ... done
Removing rails_new_redis_1 ... done
Removing network rails_new_default
Execute a command inside the app container:
$ docker-compose exec app ruby -v
ruby 2.5.1p57 (2018-03-29 revision 63029) [x86_64-linux-musl]
Execute a command inside the app container that needs env variables:
$ docker-compose exec --env RAILS_ENV=test app rails db:setup
Created database 'rails_new_test'
-- enable_extension("plpgsql")
-> 0.0251s
-- create_table("users", {:force=>:cascade})
-> 0.0366s
Adding a new gem (does not require image rebuild):
# update Gemfile
$ docker-compose exec app bundle
Running specs:
$ docker-compose exec app rspec
....................
Finished in 0.47352 seconds (files took 16.36 seconds to load)
20 examples, 0 failures
TODO
Nothing right now.
License
This project is MIT licensed, see LICENSE.