application_controller.rb

class ApplicationController < ActionController::Base
  extend AttributeViewable
end

attribute_viewer.rb

module AttributeViewer
  def attr_viewable(*names)
    names.each do |name|
      attr_accessor name
      helper_method name
      private "#{name}=".to_sym
    end
  end
end

index.html.erb

<table>
<% messages.each do |message| %>
  <tr>
    <td><%= message.subject %></td>
    <td><%= message.body %></td>
    <td><%= link_to 'Show', message %></td>
    <td><%= link_to 'Edit', edit_message_path(message) %></td>
    <td><%= link_to 'Destroy', message, confirm: 'Are you sure?', method: :delete %></td>
  </tr>
<% end %>
</table>

messages_controller.rb

class MessagesController < ApplicationController
  attr_viewable :message, :messages

  def index
    self.messages = Message.all
  end
end

Yeah, I think so. It's just a lazy way to go about things, in my opinion.

…

On Thu, Mar 22, 2012 at 2:28 PM, zobar ***@***.*** wrote: Are at-signs really that horrible? --- Reply to this email directly or view it on GitHub: https://gist.github.com/2160696

Lazy, and they invite unintended modification. An object should only have one reason (and one interface) to change.

Just thinking out loud here:

I like the idea of using methods over ivars. The thing I don't like is that the views have methods exposed to them that they shouldn't. For example, in this case, a view that's dealing with messages probably will not deal with the message method. Yet the view's context will respond to the message method and calling that method will return a nil. That means you won't know you inadvertently called that method until later down the road when you try to call a method on nil.

Maybe it would be better to set a context object. The view context would always responds to a particular method, but the type of object returned by that method changes depending on the action that's called. (I guess this is called a "presenter", though I like to call it an object that's composed of things. 😉) Then if the "messages" view inadvertently calls message on the context object, you immediately have an exception raised.

Not to mention easier testability of the composed object. :-D

@tenderlove I like how that's handled in the JS templating world, where the context is usually a javascript object. However, to apply that to this example, this.messages would be the key pointing to a collection. How else would you handle accessing a messages collection? context.each {..}?

@scottburton11 I would add a messages method to the context object. So in your view, you'd say ctx.messages.each { ... }. The context object would be constructed in each action, like this:

class MessagesController < ApplicationController
  attr_accessor :context
  helper_method :context

  IndexContainer = Struct.new(:messages)
  ShowContainer  = Struct.new(:message)

  def index
    self.context = IndexContainer.new(Message.all)
  end

  def show
    self.context = ShowContainer.new(Message.find(params[:id]))
  end
end

Then if you accidentally call context.message in the view of your index, you'll get an exception. Or if you call context.messages from your show view, you get an exception.

Ah, I see now.

How hard would it be to mixin the context methods so messages will work without the context.? I cannot currently think of an easy way to achieve that.

@myobie you could mix in a module on each request. So something like:

class MessagesController < ApplicationController
  def index
    extend Module.new {
      attr_accessor :messages
    }

    self.messages = IndexContainer.new(Message.all)
  end
end

But I strongly advise against doing that. It breaks method caches which will slow method lookup on every request, and couples your code with the fact that the controller is a new instance on every request.

Usage of a context (or container, or presenter, or whatever you want to call it) object allows you to:

• test behavior of the container outside of your controller (you don't need to write a functional test to test the Container)
• decouple the view from the controller (you could test the view without instantiating a controller)
• get NoMethodErrors raised exactly at the point where the error occurred (rather than waiting for the nil to be used)