Errata for Agile Web Development with Rails, Third Edition

#38583: When showing the regular expressions on validating, they are being misformatted in the mobipocket version.

Instead of showing as %r{(\.(gif|jpg|png)$}i They show as:

%
\.(gif|jpg|png)$}i--Joseph Erickson

#37666: "Note that before we compare the price to 0.01, we first check to see whether it’s nil. This is important: if the user leaves the price field blank, no price will be passed from the browser to our application, and the price variable won’t be set. If we tried to compare this nil value with a number, we’d get an error."

That would be true, except that you're already testing for numericality. If you leave the field blank, the form errors on the 'not a number' validation. --Ted Roche

#39054: I tried to populate the table with greek data (greek letters) but rake failed to migrate. I tried single quotes, double quotes and %{...} but no success. With english is ok.--Alexandros Kontoulis

#39207: after inputting this statement "ruby script/generate controller store index", safari wouldn't show expected page but a message saying "No route matches "/store" with {:method=>:get}". I fixed this by add "map.resources :store" in /config/routes.rb

#39315: Is not neccesary --> #39207: after inputting this statement "ruby script/generate controller store index", safari wouldn't show expected page but a message saying "No route matches "/store" with {:method=>:get}". I fixed this by add "map.resources :store" in /config/routes.rb.
Just Restart Rails webserver.--Felixaa

#39316: delete dot after "each" --> #39204: Change "for product in @products" to "@products.each.do |product|" ...
"for product in @products" to "@products.each do |product|"--felixaa

#39051: Latest rails is 2.3.2
get rails3-code.zip, then unzip.
Then go to "pragforms",
set RAILS_GEM_VERSION to 2.3.2
and tried to start the server, but could not do it.

#39124: "On line 11, we set the page heading to the value in the instance variable @page_title." Line 11 doesn't assign to @page_title. Rather, "Pragmatic Bookshelf" is displayed to the right of the image, which would imply that @page_title evaluates to nil when line 11 is executed, and a fallback string is used.

#39206: Figure 7.3: In Firefox each product has a vertical scroll bar, probably because of the #store .entry { overflow: auto; ..} Change this to "overflow: hidden;" and it works. --Lasse Bunk

#38036: Figure 7.3 shows "Add to Cart" button rendering next to the price but for me it renders below the price (Rails 2.2.2 on Safari & Firefox)
Could not figure out how to get it to render as displayed (which would be desirable)--Jay Casler

#39687: The line
config.action_controller.session_store = :active_record_store
which is supposed to be in environment.rb is no longer there. Instead, this line
ActionController::Base.session_store = :active_record_store
should be uncommented in config/initializers/session_store.rb--Reed Law

#38468: I'm running Rails 2.3.2. After uncommenting "ActionController::Base.session_store = :active_record_store" in session_store.rb, I started getting "ActionController::InvalidAuthenticityToken" error messages (as mentioned by another poster). Adding "protect_from_forgery :only => [:create, :update, :destroy]" to store_controller.rb and products_controller.rb eliminated the InvalidAuthenticityToken error when adding a product to the cart, but did not eliminate the error when creating or updating a product.

#39306: mine is the P1.0 PDF printing - Task C: Cart Creation on Sessions talks about protect_from_forgery in /app/controllers/application.rb referring to a commented out :secret parameter for the protect_from_forgery line. Version 2.3.2 of rails does not make any reference to :secret here... also as others have noted 2.3.2 also places the session_store configurations mentioned in the previous section under config/initializers/session_store.rb

Clarification on the protect_from_forgery :secret parameter would be nice here.

#39211: On page 126 you tell us that variables shouldn't be passed as globals. But now you pass @current_item as a global variable down to the cart_item partial. I can see why this could be okay because the variable isn't actually required - but is this the right way to do it?--Lasse Bunk

#38337: My code gives me an error if I simply run "redirect_to_index". I have to pass in a parameter, the error message tells me that I have to pass in a parameter. So I use "redirect_to_index(nil)" instead.

#38579: Figure 10.3: validation errors - with Rails 2.3.2 and the provided styles both the labels and the form input controls are solid red. (Checked against full depot_s downloaded code.)--Justin Forder

#37881: "We’ll create a salt string by concatenating a random number and the object
id of the user object. It doesn’t much matter what the salt is as long as it’s
unpredictable (using the time as a salt, for example, has lower entropy than
a random string). We store this new salt into the model object’s salt attribute.
Again, this is a private method, so place it after the private keyword in the
source. "

The source code for that particular method should be:

class << self
private
def encrypted_password(password, salt)
string_to_hash = password + "wibble" + salt
Digest::SHA1.hexdigest(string_to_hash)
end
end

in order for it to be a private class method. The given method in the source code still allows encrypted_password to be called publicly.
--Ronnie Liew

#38547: In the revised new.html.erb,
<%= f.label :user_password, 'Password' %>
should be
<%= f.label :password, 'Password' %>
and
<%= f.label :user_password_confirmation, 'Confirm' %>
should be
<%= f.label :password_confirmation, 'Confirm' %>

--Justin Forder

#38562: "With postback handling, there is no need to issue a redirect and therefore
no need to make flash available across requests. flash.now makes the notice
available to the template without storing it in the session." - The use of flash.now is because we are returning the user to the submitted screen (as we would with a validation failure). A successful POST to a postback handler still needs to redirect (and does, in the code given here).

--Justin Forder

#38885: The book suggests that naming a method after_destroy in the user.rb file will cause ActiveRecord to execute that method as a hook after it destroys an instance of the model.

Under my rails 2.3.2 and ruby 1.8.6, the method never gets executed.

If I change the code to this, instead, it works:

depot/app/models/user.rb

...
after_destroy :my_little_destroy_method

def my_little_destroy_method
#code raising the exception goes here
end

--Sam Aparicio

#39504: Issue presumably because of me doing something silly, or Rails 2.3.2 has changed from when the book was written.

I don't get the nice error messages when I attempt to submit a blank checkout form in Spanish. Instead, I get the following flash message displayed:

Address translation missing: es, activerecord, errors, models, order, attributes, address, blank
Name translation missing: es, activerecord, errors, models, order, attributes, name, blank
Pay type translation missing: es, activerecord, errors, models, order, attributes, pay_type, blank
Pay type translation missing: es, activerecord, errors, models, order, attributes, pay_type, inclusion
Email translation missing: es, activerecord, errors, models, order, attributes, email, blank

I can add the relevant items to the YAML file but it doesn't feel very DRY to add the same message so many times?

es:
activerecord:
errors:
models:
order:
attributes:
address:
blank: "no puede quedar en blanco"

Rails 2.3.2
$ ruby -v
ruby 1.8.7 (2008-08-11 patchlevel 72) [i486-linux]
--James Abley

#38578: With Rails 2.3.2 the ActiveRecord error messages appear to be being escaped before being displayed - so the Spanish error messages for an empty checkout form include:
- Forma de pago no est&aacute; incluido en la lista
- Direcci&oacute;n no puede quedar en blanco
(Checked with downloaded depot_s code, with the version number changed in config/environment.rb, controllers/application.rb renamed to controllers/application_controller.rb, and my existing database file copied over)

--Justin Forder

#38782: This is not an error so much as a question. I'm using mySQL and instead of getting 0 errors I get ActiveRecord::StatementInvalid: Mysql::Error: Column 'quantity' cannot be null: INSERT INTO 'line_items' (`updated_at`, `order_id`, `total_price`, `quantity`, `product_id`, `id`, `created_at`) VALUES ('2009-04-07 16:40:04', 1, 9.99, NULL, 1, 996332877, '2009-04-07 16:40:04')

#38374: It looks like "follow_redirect" was taken out from functional tests. I came across the following paragraph in the "lib/ruby/gems/1.8/gems/actionpack-2.2.2/CHANGELOG" file after running into a method_missing error during the test.

208
209 * Remove follow_redirect from controller functional tests.
210
211 If you want to follow redirects you can use integration tests. The functional test
212 version was only useful if you were using redirect_to :id=>...
213

Thanks.
--IJ

--Iljun Kim

#39056: depot> ruby -I test test/performance/order_speed_test.rb

...should be...

depot> ruby -I test test/fixtures/performance/order_speed_test.rb

...where we've nested the performance directory inside the fixtures directory. --Steven Noble

#39044: While Rails 1.2.6 assumes 30 day months, Rails 2.2.2 seems to be more accurate:

>> puts Time.now.utc
Fri May 01 03:12:51 UTC 2009
=> nil
>> puts 20.months.ago
2007-09-01 03:12:57 UTC
=> nil
--Scott Johnson

#37851: The name.chars.length gives an error:
undefined method `length' for #<Enumerable::Enumerator:0xb70f99cc>

The correct class name for multibyte characters support is mb_chars.--Dimitris Kallipolitis

#38906: On page:
named_scope :last_days, lambda{ |days| :condition => ['updated <?', days]}

Didn't work for me on rails 2.3.2 (although I assume it won't work on other versions). Worked when I changed it to

named_scope :last_days, lambda{ |days| {:conditions => ['updated <?', days]}}

ie. pluralized condition and made the hash explicit.
--Chris Causer

#38955: "Partial Updates and Dirty Bits" on p.351 says that Rails saves only the attributes that have been modified by direct assignment. But "Updating Existing Rows" on p.340 says a column will be updated in the database even if its value has not changed.

These seem to be conflicting statements. I'm not sure which is correct.
--Scott Johnson

#38980: The book says in the paragraph about counting that the count function with just one string parameter will use that parameter as a condition (just like the find function previously described). This didn't work properly when I tried. Instead (and if I understand the rails documentation right) it uses the parameter as a column name and counts all rows where the corresponding value is not NULL.--Nicolas Fesselet

#37812: Context: "user.name = 'Bill'
user.name_change # => ['Dave', 'Dave Thomas']"

This implies that the attribute change is only tracked the first time it is updated.

However, when I do this

kirk$ script/console
Loading development environment (Rails 2.1.0)
>> elt = Project.find(:first)
=> #<Project id: 1, name: "P01", organization_id: 1, created_at: "2009-02-11 17:32:43", updated_at: "2009-02-18 18:25:00">
>> elt.name = 'P1'
=> "P1"
>> elt.name_change
=> ["P01", "P1"]
>> elt.name = "P_1"
=> "P_1"
>> elt.name_change
=> ["P01", "P_1"]

it looks as if the attribute change is tracked each time you assign a new value to the attribute (although intermediate changes are lost).

--Kirk Baird

#39060: The resource_id referencing Article One should be 1 as opposed to 2. (Though the preceding page says "Let's clear out our database..." it appears this wasn't done.)--Christian Campbell

#38653: Two paragraphs contradicting each other:
:overwrite_params: "The options in hash are used to create the URL, but no default values are taken from the current request."

But above it states ":overwrite_params
option to tell url_for that the original request parameters contained the new year that we want to use. Because it thinks that the year hasn’t changed, it continues
to use the rest of the defaults."

'continue to use the rest of the defaults' != 'no default values are taken from the current request'

#39801: The second paragraph says: "... We use the pagination_links helper method to construct a nice set of links to other pages." And then the example uses 'will_paginate' instead.--Octav Popescu

#38794: In the example "create" action for saving multiple models at the bottom of page 532 and continuing on page 533 there is a "rescue" clause that has no associated "begin" which is invalid. There should be a "begin" before the line that reads "Product.transaction do".--Tres Henry

#37787: The code supplied for pragforms is still incorrect in that when used, the following error is seen:
pragforms/config/boot.rb:33:Warning: Gem::SourceIndex#search support for String patterns is deprecated
rake aborted!
undefined method `require_gem' for main:Object
Rails is 2.2.2 and gem is 1.3.1
The work around to 'rails pragforms and then copy the appropriate code to the correct folders.
Not very friendly though!!--Martin Hawkins

#39294: Is the full setup method required in the OrderControllerTest? I thought that they were no longer required with a certain improvement of rails. Also the Rails Guides does it differently by not adding the setup method and accessing the ActionMailer::Base directly rather than setting up a variable for it.--Shaun McDonald

#39619: I believe Passenger became compatible with mod_rewrite as of v2.1, and the current version is now compatible with mod_alias. Perhaps you could point people at the relevant section of the Passenger User Guide for timely information on conflicts.

#39019: The second key generation command fails for me (saying .ssh/authorized_keys2 doesn't exist). Did you mean ~/.ssh for those commands? If not, where do expect the commands to be run from? Also, it would be good if it was explained a bit more. Should the key be generated on the web server, and then copied to ~/.ssh/authorised_keys2 on the SCM server? (in my case, they are different servers).--Robert Gravina

#38377: The listing in the book for the depot.css file (p. 721) does not appear to include the logic to alternate shading when displaying the Listing Products page as shown on page 94. I cut and pasted the .css file instead of downloading it.

So far, though, excellent update. Many thanks.--Steven Rummel