ApplicationController: understanding generate scaffold

I'm new to Rails and I'm trying to understand the code created by
generate scaffold. I'm using Rails 2.2.2.

$ ./script/generate scaffold Task description:string

generated the following code in app/controller/tasks_controller.rb

class TasksController < ApplicationController
  # GET /tasks
  # GET /tasks.xml
  def index
    @tasks = Task.find(:all)

    respond_to do |format|
      format.html # index.html.erb
      format.xml { render :xml => @tasks }
    end
  end

  # GET /tasks/1
  # GET /tasks/1.xml
  def show
    @task = Task.find(params[:id])

    respond_to do |format|
      format.html # show.html.erb
      format.xml { render :xml => @task }
    end
  end

  # GET /tasks/new
  # GET /tasks/new.xml
  def new
    @task = Task.new

    respond_to do |format|
      format.html # new.html.erb
      format.xml { render :xml => @task }
    end
  end

  # GET /tasks/1/edit
  def edit
    @task = Task.find(params[:id])
  end

I noticed that the edit method is missing some code. If I remove those
same lines from each of the methods above. The code still works, why?

I couldn't find ApplicationController in the API docs at
http://api.rubyonrails.org/ ... I figured the behavior must be defined
in the super class, but I can't seem to find the code for it either.

Can someone point me in the right direction to illuminate this mystery?

Thanks in advance,
Sarah

http://www.ultrasaurus.com/code (where I'm keeping a journal of my
learning Rails so far)
http://www.ultrasaurus.com (primary blog)

The respond_to block is optional, and if the block is missing rails
will execute a render :task so for the edit task it will look for the /
controller/edit.html.erb file

Here is information from the api...
http://apidock.com/rails/ActionController/MimeResponds/InstanceMethods/respond_to

Try creating a task, and then editing it. You should find that, when you remove the lines from “Edit”, you don’t get information that you previously entered when you created the task.

If you look in app/controllers/application.rb, you will find that ApplicationController is a task that you derived from ActionController::Base. You should be able to find documentation about ActionController::Base at http://api.rubyonrails.org/

From one newbie to another…
Look at the links created in the index view for your tasks controller. With the default scaffold, you should see three links to the right of each task listed: Show, Edit, and Destroy. If you mouse-over the Edit link, you will probably see something that looks like: http://localhost:3000/tasks/1/edit. That says to the routing system in Rails (through the magic of the code you wrote in config/routes.rb) “Invoke the ‘edit’ action in the ‘tasks’ controller, and, by the way, set the ‘id’ parameter to ‘1’”.

When your #edit action is invoked, it searches the Task table in your database for the entry with an ID of 1. Then Rails, by default, will render a view (a web page) named “edit”. You can see the template for that web page in app/views/edit.html.erb.

If you completely remove the #edit action, the default behavior is still to render a template with the same name as the action.

If you leave the #edit action in there, but don’t do anything in it, the default behavior is still to render a template with the same name as the action.

If you leave the code in there as written, the #edit action will fetch the row from the Task database with the specified ID. The #edit view will then have some data to show when it displays the form.

–wpd

Patrick Doyle wrote:

If you look in app/controllers/application.rb, you will find that
ApplicationController is a task that you derived from
ActionController::Base. You should be able to find documentation about
ActionController::Base at http://api.rubyonrails.org/

ah. This solves the first mystery. Still learning all the stuff that
gets generated with scaffold.

From one newbie to another...
Look at the links created in the index view for your tasks controller.
With
the default scaffold, you should see three links to the right of each
task
listed: Show, Edit, and Destroy. If you mouse-over the Edit link, you
will
probably see something that looks like:
http://localhost:3000/tasks/1/edit.
That says to the routing system in Rails (through the magic of the code
you
wrote in config/routes.rb)

hmm. If I look at the routes that were auto-generated for me, I see:
  map.connect ':controller/:action/:id'

which would lead me to believe that the URL should be:
http://localhost:3000/tasks/edit/1

while semantically it makes more sense to apply the edit action last, I
don't see how that maps to the code. Is it considered bad practice to
leave these as is. The comment says "consider removing the them or
commenting them out if you're using named routes and resources." I'm
not quite sure what "named routes and resources" are and whether I'm
using them.

"Invoke the 'edit' action in the 'tasks'
controller, and, by the way, set the 'id' parameter to '1'".

When your #edit action is invoked, it searches the Task table in your
database for the entry with an ID of 1. Then Rails, by default, will
render
a view (a web page) named "edit". You can see the template for that web
page in app/views/edit.html.erb.

If you completely remove the #edit action, the default behavior is still
to
render a template with the same name as the action.

Yes, it does this, but when the template is rendered, it causes the
error below, which I assume is because '@task' is not defined, although
I don't really understand the error message.

Called id for nil, which would mistakenly be 4 -- if you really wanted
the id of nil, use object_id

Extracted source (around line #3):

1: <h1>Editing task</h1>
2:
3: <% form_for(@task) do |f| %>
4: <%= f.error_messages %>
5:
6: <p>

If you leave the #edit action in there, but don't do anything in it, the
default behavior is still to render a template with the same name as the
action.

yup. same behavior.

If you leave the code in there as written, the #edit action will fetch
the
row from the Task database with the specified ID. The #edit view will
then
have some data to show when it displays the form.

Makes sense.

I see empirically how it works; however, I'm still curious how does
Rails "know" whether I've called respond_to in my edit method?

Thanks so much,
Sarah

Patrick Doyle wrote:

From one newbie to another…

Look at the links created in the index view for your tasks controller.

With

the default scaffold, you should see three links to the right of each

task

listed: Show, Edit, and Destroy. If you mouse-over the Edit link, you

will

probably see something that looks like:

http://localhost:3000/tasks/1/edit.

That says to the routing system in Rails (through the magic of the code

you

wrote in config/routes.rb)

hmm. If I look at the routes that were auto-generated for me, I see:

map.connect ‘:controller/:action/:id’

which would lead me to believe that the URL should be:

http://localhost:3000/tasks/edit/1

while semantically it makes more sense to apply the edit action last, I

don’t see how that maps to the code. Is it considered bad practice to

leave these as is. The comment says "consider removing the them or

commenting them out if you’re using named routes and resources." I’m

not quite sure what “named routes and resources” are and whether I’m

using them.

Look at the top of your routes file. You will see something that looks vaguely like:

map.resources :tasks

That creates routes that match /tasks/1/edit

Oh cool… I just realized that using the RESTful routes created by map.resources, and using the default route at the end of routes file, you would get to exactly the same place via:

http://localhost:3000/tasks/1/edit

and

http://localhost:3000/tasks/edit/1

If that sort of thing bothers you, you could “consider removing [the default rules] or commenting them out if you’re using named routes and resources”

Oh, I just saw the part in your email about the “named routes and resources”. Since you are using the scaffold, you are using named routes and resources, since that is what is considered to be the best practice by those who brought you Rails. Google “RESTful routing” and you will find much more information than I could provide in an email. But, in a nutshell, the “map.resources” command at the top of the file creates a bunch of routes for you that map to the 7 actions you found in your controller. With those in place, you don’t need the two default routes at the end of the file.

"Invoke the ‘edit’ action in the ‘tasks’

controller, and, by the way, set the ‘id’ parameter to ‘1’".

When your #edit action is invoked, it searches the Task table in your

database for the entry with an ID of 1. Then Rails, by default, will

render

a view (a web page) named “edit”. You can see the template for that web

page in app/views/edit.html.erb.

If you completely remove the #edit action, the default behavior is still

to

render a template with the same name as the action.

Yes, it does this, but when the template is rendered, it causes the

error below, which I assume is because ‘@task’ is not defined, although

I don’t really understand the error message.
Oops, I thought about that after I sent my email. But by then I figured you would have noticed it yourself :slight_smile:

The error message is produced because @task evaluates to nil. Something somewhere in the #form_for helper tried to call @task.id, which got evaluated as nil.id, which triggered an exception, which displayed an error message indicating that you probably didn’t want to do that.

I see empirically how it works; however, I’m still curious how does

Rails “know” whether I’ve called respond_to in my edit method?

The respond_to stuff has to do with “Web Services” whatever those are. (Keep in mind, I’m a newbie here too.) From what I’ve intuited so far, web services seem to like to exchange data using XML. I don’t really know why #edit doesn’t include a call to respond_to where each of the other 6 actions do, except to note the use of #edit in the world of RESTful routing. In that world, when you want to create a new record in a table, you are first presented with a view of a blank record (via the #new action). When you click on the “Create” button, the #create action gets invoked. In the same vein, when you want to edit an existing record, you are first presented with a view of the existing record (via the #edit action). When you click on the “Update” button, the #update action gets invoked. My guess is that, in the web services world, one would never invoke an “update” action without first having some idea of what the data looked like and that one would probably have learned that via the #show action, so that rendering something in XML for #edit is not necessary. In the web brower world, the “show” view displays data on a page while the “edit” view would presumably display a form (with the exact same data) allowing the end user to change the data and submit an update.

That’s my guess anyway.

–wpd