packs-rails
establishes and implements a set of conventions for splitting up large monoliths built on top of the packs
standard. packs-rails
makes it easy to use packwerk
to modularize your rails app. With packs-rails
, new packages' autoload paths are automatically added to Rails, so your code will immediately become usable and loadable without additional configuration.
Here is an example application that uses packs-rails
:
package.yml # root level pack
app/ # Unpackaged code
models/
...
packs/
my_domain/
package.yml # See the packwerk docs for more info
package_todo.yml # See the packwerk docs for more info
app/
public/ # Recommended location for public API
my_domain.rb # creates the primary namespaces
my_domain/
my_subdomain.rb
services/ # Private services
my_domain/
some_private_class.rb
models/ # Private models
some_other_non_namespaced_private_model.rb # this works too
my_domain/
my_private_namespaced_model.rb
controllers/
views/
config/
initializers/ # Initializers can live in packs and load as expected
lib/
tasks/
spec/ # With packs-rails, specs for a pack live next to the pack
public/
my_domain_spec.rb
my_domain/
my_subdomain_spec.rb
services/
my_domain/
some_private_class_spec.rb
models/
some_other_non_namespaced_private_model_spec.rb
my_domain/
my_private_namespaced_model_spec.rb
factories/ # packs-rails will automatically load pack factories into FactoryBot, see Ecosystem and Integrations#factory_bot below
my_domain/
my_private_namespaced_model_factory.rb
my_other_domain/
... # other packs have a similar structure
my_other_other_domain/
...
Setting up packs-rails
is straightforward. Simply by including packs-rails
in your Gemfile
in all environments, packs-rails
will automatically hook into and configure Rails.
From there, you can create a ./packs
folder and structure it using the conventions listed above.
If you wish to use a different directory name, eg components
instead of packs
, you can customize this by configuring packs.yml
. See packs
for more information.
If you want to have your packs code namespaced without adding a directory with the pack's name into the directory structure, you can use the automatic namespaces gem.
If you use spring, when you move a pack, you'll want spring to update its cache
Add this to one of your initializers
Packs::Specification::Configuration.fetch.pack_paths.each do |dir|
Dir["#{dir}/package.yml"].each do |package_yml|
Spring.watch(package_yml)
end
end
packs-rails
allows you to split your application routes for every pack. You just have to create a file describing your routes and then draw
them in your root config/routes.rb
file (NOTE: the draw
function is only in Rails 6.1+).
# packs/my_domain/config/routes/my_domain.rb
resources :my_resource
# config/routes.rb
Rails.application.routes.draw do
draw(:my_domain)
end
You can add engine: true
to your package.yml
to make it an actual Rails engine.
It implicitly sets up a namespace for your pack and adds namespace isolation.
You can then also use other functionality that engines provide, like per-pack routes, scoped url helpers, or a table name prefix.
# packs/my_pack/package.yml
enforce_dependencies: true
enforce_privacy: true
metadata:
engine: true
Add engine_name:
to your package.yml
to use a specific, maybe namespaced, engine name instead of the last package folder name.
# packs/my_pack/package.yml
enforce_dependencies: true
enforce_privacy: true
metadata:
engine: true
engine_name: namespaced/my_pack
The engine is created as Namespaced::MyPack::Engine
instead of MyPack::Engine
.
Simply add --require packs/rails/rspec
to your .rspec
.
Or, if you'd like, pass it as an argument to rspec
:
$ rspec --require packs/rails/rspec ...
Integration will allow you to run tests as such:
# Run all specs in your entire application (packs and all):
rspec
# Run just that one test:
rspec spec/some/specific_spec.rb
# Run all tests under the "foobar" pack and all the tests of its nested packs:
rspec packs/foobar
# Same as above but also adds the "binbaz" pack:
rspec packs/foobar pack/binbaz
# Run all test files inside the "packs/foobar/spec" directory:
rspec packs/foobar/spec
# Run all specs under the "packs/foobar/nested_pack" pack:
rspec packs/foobar/nested_pack
Ensure you have the gem in your Gemfile, and that require: false
is not present, otherwise the packs-rails
FactoryBot integration will not be run upon application/environment load.
This integration will automatically add the [spec|test]/factories
directory from each pack to the application's config.factory_bot.definition_file_paths
.
Using the example application above:
...
packs/
my_domain/
...
spec/ # OR test/
...
factories/
my_domain/
my_private_namespaced_model_factory.rb
my_other_domain/
...
spec/ # OR test/
...
factories/
my_other_domain/
my_private_namespaced_model_factory.rb
rubocop-rails targets paths starting with app/
. You might need to override the Include
globs to include your packs code.
Examples:
Rails/EnumHash:
Include:
- '**/app/models/**/*.rb'
Rails/InverseOf:
Include:
- '**/app/models/**/*.rb'
Rails/LexicallyScopedActionFilter:
Include:
- '**/app/controllers/**/*.rb'
- '**/app/mailers/**/*.rb'
parallel_tests
has it its own spec discovery mechanism, so packs-rails's RSpec integration doesn't do anything when you use them together.
To make them work, you'll need to explicitly specify the spec paths:
RAILS_ENV=test bundle exec parallel_test spec packs/**/spec -t rspec <other_options>
Similarly, Knapsack (and its Pro version) has its own spec discovery mechanism and the API will find and queue the relevant specs. To make it discover your pack tests as well, you'll need to configure the following variables:
KNAPSACK_PRO_TEST_DIR: spec
KNAPSACK_PRO_TEST_FILE_PATTERN: '{spec,packs}/**{,/*/**}/*_spec.rb'
Setting KNAPSACK_PRO_TEST_FILE_PATTERN
will tell Knapsack where your specs are located, while setting KNAPSACK_PRO_TEST_DIR
is necessary because otherwise an incorrect value is sent to rspec via the --default-path
option.
Bug reports and pull requests are welcome on GitHub at https://github.com/rubyatscale/packs-rails.