domain and persistence

01-domain_trip.rb

class Domain::Trip
  include Virtus
  include ActiveModel::Validations
  include ActiveModel::Conversion
  extend  ActiveModel::Naming

  attribute :id
  attribute :name
  attribute :sections, Array[Section], :default => []
end

02-persistence_trip.rb

class Persistence::Trip < ActiveRecord::Base
  def self.find_by_id(id)
    record = super(id)
    Domain::Trip.new(record.attributes).tap do |trip|
      trip.sections = record.sections.map { |section_record| Domain::Trip::Section.new(section_record.attributes) }
    end
  end
end

03-trips_controller.rb

def show
  @trip = Persistence::Trip.find(params[:id)
end

04-index.html.haml

- @trip.sections.each do |section|
  = section.name

You can give Virtus a nested hash, and it will hydrate the chid objects in the graph automatically.

I'm not sure if you can get AR to give you a nested hash of attributes out of the box, but it would be trivial to add that.

I've played with Virtus and nested hashes, AR#attributes does not return a deep hash, but as you say it could easily be done. I think my problem is more do with the API of the persistence class - what should the different calls to the persistence class look like. If say I want to fetch a trip and all its associations (e.g show action) versus just the trip (e.g index action).

I think you probably have two different value objects there then: a TripSummary for the index page, with a shallow cut of the data, and a Trip with the deep set of data. I'd just put the queries for those behind two separate methods.

Thanks - I like the idea of having TripSummary, after all it does not need to contain any validations, or anything else that acts on state changes, its being used in a read-only context. I guess I need to get over fear of adding classes which have overlapping responsibilities, its marginally less DRY but simpler than trying cram everything in to one class.