Dry-web-roda for Rails Developers Part I

Revised by Piotr Solnica and Andy Holland

Lately, I have been playing around and contributing to the great ecosystem of dry-rb, first of all, I have to say the community is absolutely fantastic, super supportive and eager to welcome many new contributors.

First I will like to thank Piotr Solnica, Andy Holland, Tim Riley and Nikita Shilnikov - they have been really helpful and patient with my many questions.

Yesterday I decided to start playing around with a gem call dry-web-roda this small framework aims to provide an alternative to building web apps using ruby, with the use of small libraries such as dry-view, dry-container, dry-transaction, roda, rom and many more; the help you build clearer, flexible and more maintainable code.

The post tries to create a bridge for Rails Developers and encourage them to try these new alternatives, that will bring joy and fresh concepts for building web apps with ruby.

After installing the gem we can create a sample app by typing dry-web-roda new github_stalker --arch=flat- this will create the file structure.

├── bin
├── config
├── db
├── lib
│   ├── github_stalker
│   │   └── views
│   └── persistence
│       ├── commands
│       └── relations
├── log
├── spec
│   └── support
│       └── db
├── system
│   ├── boot
│   └── github_stalker
├── transactions
└── web
    ├── routes
    └── templates
        └── layouts

At first the structure is quite different from what we are used to in a typical Rails app, but I will try my best to explain it.

First, the system folder is where all the configuration lives, we can think of them as our initializers, this will be the entry point of our application. There are many small libraries involved to make everything work, that I can not go into detail in just one post, I will try cover all of them in a series of posts. ├── system    ├── boot    │   ├── monitor.rb    │   ├── rom.rb    │   └── view.rb    ├── boot.rb    └── github_stalker    ├── application.rb    ├── container.rb    ├── import.rb    ├── repository.rb    ├── settings.rb    ├── transactions.rb    ├── view_context.rb    └── view_controller.rb We will start with application.rb, this file contains all the configuration regarding routes, container and plugins to be use with roda.

require "dry/web/roda/application"
require_relative "container"

module GithubStalker
  class Application < Dry::Web::Roda::Application
    configure do |config|
      config.container = Container
      config.routes = "web/routes".freeze

    opts[:root] = Pathname(__FILE__).join("../..").realpath.dirname

    use Rack::Session::Cookie, key: "github_stalker.session", secret: GithubStalker::Container.settings.session_secret

    plugin :csrf, raise: true
    plugin :flash
    plugin :dry_view

    route do |r|

      r.root do
        r.view "welcome"


What is a container? I’m going to bring the words from the official website dry-container is a simple, thread-safe container, intended to be one half of the implementation of an IoC container or how I understand it a container “gives you access to the objects that make up your application”.

Roda (the router) is a Routing Tree Web Toolkit I will not go into much detail since I’m really new to it.

Following is the container.rb and import.rb, which in my opinion is where all the magic happens, thanks to dry-container and dry-auto_inject. This holds the configuration for loading the files for our application, more or less like auto_load_path of Rails, but only using ruby methods and variables require and $LOAD_PATH.

require "dry/web/umbrella"
require_relative "settings"

module GithubStalker
  class Container < Dry::Web::Umbrella
    configure do
      config.name = :github_stalker
      config.default_namespace = "github_stalker"
      config.settings_loader = GithubStalker::Settings
      config.listeners = true

      config.auto_register = %w[

    load_paths! "lib", "system"

    def self.settings

At first sight, we see some configuration for name, namespace and some auto_register, this will register all the files inside our lib/github_stalker folder in your container, following the convention from the file structure, so for example

├── lib
    ├── github_stalker
        ├── github
        │   ├── client.rb # will register inside our container under the name github.client
        │   ├── fetch_gists.rb # github.fetch_gists
        │   └── fetch_info.rb # github.fetch_info
        └── users
            └── validate_input.rb # users.validate_input

And thanks to dry-auto_inject, all these files will be lazily loaded as required, making it really efficient.

require "github_stalker/import"

module GithubStalker
  module Github
    class FetchGists
      include GithubStalker::Import['github.client'] # at the instance level we will have access to `client`

      def call(input)

Well I think this post is getting too long and I don’t want to take more of your time. Thank you for reading it. I will keep creating more posts explaining the rest of the structure and libraries involved in dry-rb - they are great and bring a fresh view in the ruby world.

All the above example were taken from an example app I built using dry-web-roda if you want to check the code please follow this repo

If you have any thoughts or questions, please share and I’ll be happy to answer in the comments.

comments powered by Disqus