Rails Cookbook
by Rob Orsini

The unconfirmed error reports are from readers.  They have not yet been 
approved or disproved by the author or editor and represent solely the
opinion of the reader.

Here's a key to the markup:
        [page-number]: serious technical mistake
        {page-number}: minor technical mistake
        <page-number>: important language/formatting problem
        (page-number): language change or minor formatting problem
        ?page-number?: reader question or request for clarification

This page was updated April 14, 2008.


UNCONFIRMED  errors and comments from readers: 

=======================

{00} Recipe 3.16. Modeling a Threaded Forum with acts_as_nested_set;
It would be DRYer with better encapsulation in this recipe to move some of the logic to the post model 
with a method like "is_viewable?", keeping the logic in one place instead of in the helper and view.

Also you don't need to use a global variable in your recursive helper, and you need a block element (like 
<p>) to wrap each post if you want it to display as you have it in the screenshot.

model:

class Post < ActiveRecord::Base
acts_as_nested_set

def is_viewable?
send(parent_column) != nil
end
end

helper:

module PostsHelper
def get_indentation(post, n=0)
return n unless post.is_viewable?
parent = Post.find(post.send(post.parent_column))
get_indentation(parent, n += 1)
end
end

view:

<h1>Threaded Forum</h1>

<% for post in @posts %>
<p>
<% get_indentation(post).times do %>_&nbsp;<% end %>
<%= post.subject %>
<i>[
<% if post.is_viewable? %>
<%= link_to "view", :action => "view", :post => post.id %> |
<% end %>
<%= link_to "reply", :action => "new", :parent => post.id %>
]</i>
</p>
<% end %>


[00] Recipe 3.18. Avoiding Race Conditions with Optimistic Locking;
First sentence after Solution says:

"There is no way to force Rails to lock a row for later update. This is commonly known as pessimistic 
locking or select for update. To lock a row with Active Record, you need to use optimistic locking."

However, Rails does support pessimistic locking. See the API at:

http://api.rubyonrails.com/classes/ActiveRecord/Locking/Pessimistic.html

"Locking::Pessimistic provides support for row-level locking using SELECT  FOR UPDATE and other lock 
types."


(00) Recipe 3.22. Mixing Join Models and Polymorphism for Flexible Data Modeling;
The second code example for magazine.rb says:

has_many :subscribers, :through => :subscriptions

but it should be:

has_many :readers, :through => :subscriptions 


{00} Recipe 4.15. Tracking Information with Sessions;
The square root of 4 is 2, not 16. Maybe you meant the 4 squared?

{ :question => "What's the square root of 4?",
:options => ['16','2','8'],
:answer => '2' },


{00} Recipe 4.16. Using Filters for Authentication;
The UsersController needs to require the 'digest/sha1' module for calls to Digest::SHA1.hexdigest.

app/controllers/users_controller.rb:

require 'digest/sha1'

class UsersController < ApplicationController

{00} 5.11. Processing Dynamically Created Input Fields;
The order of user.id and role.id is switched in the helper get_perm method, resulting in the controller 
not being able to update the permissions table.

user_helper.rb:

def get_perm(role_id, user_id)

should be:

def get_perm(user_id, role_id)


(21)  second paragraph from the bottom;
The line:
> ruby script\generate migration add_photos
does not work for me ( lots of lines of errors ).
So I tried:
> rails script/generate migration add_photos
and it appears to work.

But then I got confused because the next bit says:
edit the file db/migrate/add_photos.rb

But it should read:
edit the file db/migrate/001_create_photos.rb

add_photos is the class within this .rb file


(27) First full paragraph after the call app/controllers/language_controller.rb;
First sentennce of the paragraph reads:
Here, you are passing the scaffold method a symbol representing your model; :languages in this case.

It should match the code just above it: scaffold :language.

So the sentence should read:
Here, you are passing the scaffold method a symbol representing your model; :language in this case.


(43) Last paragraph (between headings and bulleted list);
The descriptions of common RDoc formatting options should precede the corresponding examples, rather than 
trail them. E.g.,

Here are some other common formatting options:
# = Heading One
#
# == Heading Two
#
# === Heading Three
The following produces heading [sic] of various sizes:

should be changed to something like:

Here are some other common formatting options--the following produces headings of various sizes:
# = Heading One
#
# == Heading Two
#
# === Heading Three

etc. The same problem is repeated for all four examples on pp. 43-44.



======================
{50} 0th paragraph;
This is from the January 2007 Printing.

From text:
"For example, a students table referencing another table named courses would contain a courses_id column"

This should read:
"...contain a course_id column"

Note that the name of the foreign key field is in the singular.


[51] Create table recipes and create table recipes_tags are not the same for mysql and postgres; the mysql definition is primary key of id, chapter_id, and title. the postgres definition is primary key of id.

The mysql table is defined like this in the book:

create table recipes (
id int not null auto_increment,
chapter_id int not null,
title varchar(255) not null,
problem text not null,
solution text not null,
discussion text not null,
see_also text null,
sort_order int not null default 0,
primary key (id, chapter_id, title),
foreign key (chapter_id) references chapters(id)
)
type=innodb;

As such, would allow id to have many reuses of id where as the postgres version only allows id to occur 1 time in the table. I am assuming that the postgres version is what was intended and thus the mysql table would be defined as this:

create table recipes (
id int not null auto_increment,
chapter_id int not null,
title varchar(255) not null,
problem text not null,
solution text not null,
discussion text not null,
see_also text null,
sort_order int not null default 0,
primary key (id),
foreign key (chapter_id) references chapters(id)
)
type=innodb;

Also, the table recipes_tags for mysql would allow a recipe to have many tags and a tag to have many recipes... where as the postgres version has several flaws. It is defined in the book as this:

create table recipes_tags (
recipe_id serial unique references recipes(id),
tag_id serial unique references tags(id)
);

Serial should not be used for a column that is being used for reference to another table. The first reason being that it creates a default of nextval for a column that is supposed to be using an id from another table. It should be integer. This insert would be valid given the serial definition for the colunm:

insert into recipes_tags (recipe_id)values(1);

as long as there was a recipe of id 1 and a tag of the id generated by the sequence.

Also the unique definition is not the same for the mysql table recipes_tags and the postgres version. placing unique on each of the columns in the postgres version causes the table to only allow a recipe_id to be entered one time and likewise for the tag_id.

Thus to be equivalent to the mysql table defined as this:

create table recipes_tags (
recipe_id int not null,
tag_id int not null,
primary key (recipe_id, tag_id),
foreign key (recipe_id) references recipes(id),
foreign key (tag_id) references tags(id)
)
type=innodb;

I would suggest this definition for postgres:

create table recipes_tags (
recipe_id integer not null references recipes(id),
tag_id integer not null references tags(id),
primary key (recipe_id, tag_id)
);

These were tested against mysql 5.0.45 and postgres 8.2.4.


{77} Recipe 3.10, Updating an Active Record Object;
Since the book model specifies that you can have multiple inserts and these should be able to change, a 
better solution would be to use the following multiple select, or better yet a set of checkboxes, and 
allow ActiveRecord's update_attributes handle the model association updates for you:

edit.rhtml:

<select name="book[insert_ids][]" multiple="multiple">
<% for insert in @inserts %>
<option value="<%= insert.id %>" <% if @book.inserts.include?(insert) %> selected="selected" <% end %>><
%= insert.name %></option>
<% end %>
</select>

or better...

<ul>
<% @inserts.each do |insert| %>
<li><input type="checkbox" name="book[insert_ids][]"
value="<%= insert.id %>"
<% if @book.inserts.include?(insert) %> checked="checked" <% end %> />
<%= insert.name %></li>
<% end %>
</ul>

books_controller.rb:
(Less code and fewer calls to the database.)

def update
@book = Book.find(params[:id])
if @book.update_attributes(params[:book])
flash[:notice] = 'Book was successfully updated.'
redirect_to :action => 'show', :id => @book
else
render :action => 'edit'
end
end

=======================
{84-85} Recipe 3.12. Executing Custom Queries with find_by_sql;
For the migration, if you explicitly set the genre_id for the movie using Movie.create, you need to pass 
in the genre id explicitly, not just the genre object, when you create the movie. Otherwise the movie 
will default to a genre_id of 1.

db/migrate/001_build_db.rb:

...

Movie.create :genre_id => genre1.id,
:name => 'Mishi Kobe Niku',
:sales => 234303.32,
:released_on => '2006-11-01'

...

Alternatively, and I think this is probably what you meant to do, you can set the genre attribute upon 
creation instead of explicitly setting the genre_id:

Movie.create :genre => genre1,
:name => 'Mishi Kobe Niku',
:sales => 234303.32,
:released_on => '2006-11-01'


=======================


{92} Recipe 3.14. Adding Sort Capabilities to a Model with acts_as_list;
Since you've defined order as position for the book model, then you don't need to do a separate lookup 
for @chapters in the controller, and instead you can reference @book.chapters from the view and they will 
be in the correct order (by position).

Also, if you keep track of the chapters size in the controller, you can avoid extra calls to the database 
for chapter.first? and chapter.last?.

Here's the code I'd recommend, with the old code commented out for comparison:

controller:

def list
@book = Book.find(params[:id])
# @chapters = Chapter.find(:all,
# :conditions => ["book_id = %d", params[:id]],
# :order => "position")
@chapters_size = @book.chapters.size
end

view:

<h1><%= @book.name %> Contents:</h1>

<ol>
<%# for chapter in @chapters %>
<% for chapter in @book.chapters %>
<li>
<%= chapter.name %>
<i>[ move:
<%# unless chapter.first? %>
<% unless chapter.position == 1 %>
<%= link_to "up", { :action => "move",
:method => "move_higher",
:id => params["id"],
:ch_id => chapter.id } %>

<%= link_to "top", { :action => "move",
:method => "move_to_top",
:id => params["id"],
:ch_id => chapter.id } %>
<% end %>

<%# unless chapter.last? %>
<% unless chapter.position == @chapters_size %>
<%= link_to "down", { :action => "move",
:method => "move_lower",
:id => params["id"],
:ch_id => chapter.id } %>

<%= link_to "bottom", { :action => "move",
:method => "move_to_bottom",
:id => params["id"],
:ch_id => chapter.id } %>
<% end %>
]</i>
</li>
<% end %>
</ol>

=======================

{100} Recipe 3.16. Modeling a Threaded Forum with acts_as_nested_set;
It would be DRYer with better encapsulation in this recipe to move some of the logic to the post model 
with a method like "is_viewable?", keeping the logic in one place instead of in the helper and view.

Also you don't need to use a global variable in your recursive helper, and you need a block element (like 
<p>) to wrap each post if you want it to display as you have it in the screenshot.

model:

class Post < ActiveRecord::Base
acts_as_nested_set

def is_viewable?
send(parent_column) != nil
end
end

helper:

module PostsHelper
def get_indentation(post, n=0)
return n unless post.is_viewable?
parent = Post.find(post.send(post.parent_column))
get_indentation(parent, n += 1)
end
end

view:

<h1>Threaded Forum</h1>

<% for post in @posts %>
<p>
<% get_indentation(post).times do %>_&nbsp;<% end %>
<%= post.subject %>
<i>[
<% if post.is_viewable? %>
<%= link_to "view", :action => "view", :post => post.id %> |
<% end %>
<%= link_to "reply", :action => "new", :parent => post.id %>
]</i>
</p>
<% end %>


=======================

[107] Recipe 3.18. Avoiding Race Conditions with Optimistic Locking;
First sentence after Solution says:

"There is no way to force Rails to lock a row for later update. This is commonly known as pessimistic 
locking or select for update. To lock a row with Active Record, you need to use optimistic locking."

However, Rails does support pessimistic locking. See the API at:

http://api.rubyonrails.com/classes/ActiveRecord/Locking/Pessimistic.html

"Locking::Pessimistic provides support for row-level locking using SELECT  FOR UPDATE and other lock 
types."

=======================

{114} bottom example code;
Has a mistake were is write ":type =>" the correct is ":location =>" . The "type" field doesn't exist on 
table "phone_numbers".

change:

:type => "fake")

and

:type => "Fake office line")

to:

:location => "fake")

and

:location => "Fake office line")

======================

(115) Recipe 3.22. Mixing Join Models and Polymorphism for Flexible Data Modeling;
The second code example for magazine.rb says:

has_many :subscribers, :through => :subscriptions

but it should be:

has_many :readers, :through => :subscriptions 

======================

{130} First sentence at top of page;
"The form submits to the Check controller..." should be
"The form submits to the check action..."


{130} Bottom;
"...and his password is displayed...": The success.rhtml code doesn't display the password (in 
params[:pass]).


{142} Discussion section for Recipe 4.12;
The first paragraph of the Discussion states that "Protected methods can be invoked by other objects of 
the same class and its subclasses. Private methods can be invoked only by an object on itself."

Yet in the third paragraph of the Discussion, the private method double_bonus is stated to be 
"...available only to other methods in the Employee controller or its subclasses."

If double_bonus is private, it is not available to subclasses of the Employee controller (according to 
the first paragraph).


[257] test/unit/book_test_crud.rb example;
If the book model contains a validates_uniqueness_of check on :isbn (it *should*) or :title (it *could*), 
then the first assertion:

assert lisp_cookbook.save

will fail, since it's trying to save a new record in the db with the same title and ISBN as the first 
record in the books.yml fixtures file.

I think the first operation in the test_book_CRUD method should be creating a new, unique book--not one 
that already exists--then do the rest of the checks against the new book.



{279} 3rd 4th paragraph;


<% form_tag({ :action => "add" },
:id => id, :enctype =>
"multipart/form-data") do %>

recommended
:id => object_id

to avoid the warning when you run mongrel_rails start -d


(279) In code index.rhtml;
<%= link_to_remote "Add field",
:update => "files",
:url => { :action => "add_field" },
:position => "after" %> ; <-- Extra ;


[278] Adding DOM;
Does not work in IE.. does work in Firefox



[280] 2nd 3rd paragraph;
undefined method `full_orginal_filename' for #<File:C:/DOCUME~1/shay/LOCALS~1/Temp/CGI828.0>


[281] Below the datbase schema at the top of the page;
This example gives you the database schema to be used by the example from db/schema.rb, but fails to 
mention you must "ruby script\generate model customer" after the database is created. Without the model 
generated the example code will throw an uninitialized constant exception when you execute.

In the file report.rhtml around line #3

<% for column in Customer.column_names %>

Customer.column_names will throw the exception because the Customer object will not be instantiated when 
called without the model generated.



[306] 4th paragraph;
NameError in Search#index

Showing app/views/search/_search_results.rhtml where line #2 raised:

undefined local variable or method `article_body' for #<#<Class:0x3e27fa4>:0x3e27f7c>

Extracted source (around line #2):

1: <p>
2: <%= highlight(article_body, search_text,
3: '<a href="http://en.wikipedia.org?search=\1" id="results"
4: title="Search Wikipedia for \1">\1</a>') %>
5: </p>

Trace of template inclusion: /app/views/search/index.rhtml

RAILS_ROOT: ./script/../config/..
Application Trace | Framework Trace | Full Trace

#{RAILS_ROOT}/app/views/search/_search_results.rhtml:2:in 
`_run_rhtml_47app47views47search47_search_results46rhtml'
#{RAILS_ROOT}/app/views/search/index.rhtml:11:in `_run_rhtml_47app47views47search47index46rhtml'



it happens the first time you access the search_controller


(308) db/migrate/001_create_terms.rb;
The Term.create entry for the definition of ISDN has a "v" before existing. It reads: Basically a way to 
move data over vexisting regular phone lines.


(322) First paragraph in Solution section;
ActionMailer::Base.server_settings = {...}

has been renamed to

ActionMailer::Base.smtp_settings = {...}


[353]  6th paragraph;
The change to edit.rhtml should be (and it took me a day to figure this out since Im
such a newbie)

<%= start_form_tag({:action => 'update', :id => @user},  :multipart => true) %>

You need to get the :id in there or the standard update method in the controler can't
figure out how to do the @user = User.find(params[:id])

This assumes that one is using the scaffolding of rails to generate the controller.


{373} 1st paragraph;
att_accessible should be attr_accessible


{404} middle of the page;
This line is missing the destination directory for the cp command:
$ sudo cp /usr/lib/ruby/gems/1.8/gems/\
>mongrel_cluster-0.2.0/resources/mongrel_cluster

It should read:
$ sudo cp /usr/lib/ruby/gems/1.8/gems/\
>mongrel_cluster-0.2.0/resources/mongrel_cluster /etc/init.d/


[428] In config/mongrel_cluster.yml;

Page 428: (In config/mongrel_cluster.yml):
cwd: /var/www/cookbook/current
should be:
cwd: /var/www/apps/cookbook/current

Everywhere else the recipe refers to the app being located at /var/www/apps/#{application}


[456] acts_as_dictionary/init.rb code;
It's entirely possible I'm wrong, as I'm fairly new to Rails, but the line

ActiveRecord::Base.send(:include,ActiveRecord::Acts::Reportable)

should (?) read

ActiveRecord::Base.send(:include,Cookbook::Acts::Reportable)

when I used the published line I got

`load_missing_constant': uninitialized constant ActiveRecord::Acts::Reportable (NameError)

when starting mongrel

mongrel starts fine if I do the replacement above, using my own namespace


{461} second example;
The example
ActionView::Base.send :include, XhtmlValidationHelper
should read
ActionView::Base.send :include, W3cValidationHelper
or webrick will not start


{474} Discussion paragraph;
FYI, this recipe does *NOT* work with Rails 1.2. Please see the plugin author's comments:

http://www.rails-engines.org/news/2007/01/23/farewell-login_engine-/

It should be clearly noted that Rails 1.2 does not support this recipe since most users will be using 1.2. (It was beyond frustrating spending hours trying to figure out why the recipe did not work on any of my systems!)

======================