Errata for Agile Web Development with Rails, Third Edition

#32613: Hi,

As I really don’t know the answer to this question, I am not certain how relevant or desireable this would be but I would like to see a section on database modelling tools.

I am in the process of writing a spec for a new Rails app and it would be nice to have a section that briefly covered database modelling tools, that may fit nicely with the Rails environment.

I am at the point where I am writing a database Schema specification and would love to know if there is anything that would be advantageous to use for this process, and if not then it may be worth a mention that this is the case.

I appreciate that rails development is all about iteration and not so much planning but I have a major application that needs a visual representation of the database and or model and really needs plannoing in advance and it might fit in nicely with the Rails Is Agile 1.1 section.

I have got to the stage where this project already involves 20 tables and is growing every day and as I am my own customer as such it would be really good. Honest :-)

Just a thought

James--James West

#32007: Add the Rails version into the install command:
"sudo gem install -v=2.0.2 rails"
Using just "sudo gem install rails" will install a newer release - making new funny things not descriped in the book.
See page 36, 40 (rails _2.0.2_ dummy_app), 43, ... also.--wingfire

#32163: You may want to mention that to install SQLite as a RubyGem in Windows, not only must you use the 'gem' command, but you must also manually download sqlite3.dll from the SQLite site and place it in your path (e.g. /rails/bin/). Otherwise, the rails web app will return a very unhelpful "Unknown error."--Daniel Wilhelm

#33074: I am trying this on a linux system that uses postgresql rather than sqlite3. When trying to perform the say/hello action to get the Missing Template screen, I got several errors about database. To get around it, I had to modify the config/database.yml file and issue postgres commands to create the database. Perhaps a Note here would help those who aren't using exactly the same setup would be useful.--Bill Colvin

#32165: You may want to be less flippant when answering how the view gets the time in the "Joe Asks" box. Rather than saying "magic" (hence making reading the box a waste of time), some readers (such as myself) want a more advanced understanding of how this works.--Daniel Wilhelm

#31903: Could you also mention socket connections? Typical Mac OS X dev platform for many users is MAMP, which uses sockets instead of host. Maybe this is outside the scope of the document.--Ben Love

#33022: In the commandline examples: It is sometimes hard to distinguish what is to be typed and what is actually output from the commandline (especially with line wraps). I suggest to clear this that a different colour or bold is used for the command itself.--Egil Helland

#32305: In the example we explicitly add ":scale => 2" to store prices with two digits after the decimal. I am following along using sqlite3 as the back end, and the app seems to store at least four digits after the decimal. Perhaps sqlite does not use this directive? If so, the reader might be warned, or a different example chosen.--JRB

#32403: I would like a better explanation of the version number in the database. I can see that part of the version number is the date but the rest is gibberish to me. This is what one of mine looks like:
20080617021145--Ron Green

#32304: It is not clear why, in the addition to new.html.erb, the "Price" label is added using literal HTML, whereas in the already existing template code, and in the following edit.html.erb modification, you use the f.label function vie ERb.--JRB

#31999: Maybe I'm slow, but last paragraph talks about "Product model class" and where the method should be placed. I think at this stage you should still show the file and what it will look like. OK looking ahead I see on page 86 you do that. Then you should say that a number of changes are going to be made to app/models/product.rb and the final result will be shown on pg. 86.

Second suggestion: Make "Download foo" font larger.
The "Download foo" shwon in SMALL type and with a GRAY background is hard to read on the screen. Probably fine in printed book. I'm following along in my app and want to look at the existing file.

--Greg

#32583: When you start talking about validates_presence_of, it would be nice, in the margin, to remind the reader that this is an example of a declaration, and link to appendix A.4, on p. 651, where this concept is defined.

I looked at it and was momentarily puzzled, coming from a C/C++/Java background, as to what, if anything, the statement would do.--Michael Rogers

#31911: "Note also that, because it uses Rails’ create method, it will fail silently if records cannot be inserted due to validation errors." -> This note would be clearer if you had already explained the methods available on models, and their relationship to validation. Alternatively, link somewhere that does explain that.--Antony Blakey

#32249: Just to clean up the HTML a bit, and separate presentation from content a little better, instead of

<table cellpadding="5" cellspacing="0">
<tr valign="top" class="....">

Put in for HTML:

<table>
<tr class="....">

And then in the CSS:

table { border-collapse: collapse; padding: 5px; }

Which does the same thing as the attributes in your table tag, and add to existing style declarations for classes list-line-odd and list-line-even in your stylesheet:

.list-line-odd, .list-line-even { vertical-align: top; }

(and you can remove the valign attribute in the tr tag). You could, optionally, move the "width=60%" attribute over to the stylesheet too, but what I'd do (just to save on code), is use the class "list-actions" and declare that as width=40% in your stylesheet, just so that you don't have to declare a new class.

Also, for the title/description pairing, instead of using a <span> on the title, it *might* be a little more semantic/accessible to use a definition list <dl>, and to pair the title <dt> and the description <dd> with each other. That's the way I would do it if I were creating the site for a client, but it's more a matter of personal style rather than a better way to do it.

Feel free to contact me if you have any questions: jonathan.lane@gmail.com--Jonathan Lane

#32420: after adding the storecontroller, i received an error "No route matches "/login/index" with {:method=>:get}".

restarting mongrel cleared the problem.

might be providing this as a hint.
why do i need to restart mongrel is a secondary question that the authors might like to shed some light on.

i am running 2.0--mike h

#33093: Please emphasize this code: "def self.find_products_for_sale
find(:all, :order => "title")
end" has to go before the validation lines to avoid an error...--Jose Luis Velasco

#32484: I like the yellow arrows showing what new code to add to previous code. I think it should belong on page 102 to illustrate to the reader to add this code:

<%= button_to "Add to Cart" , :action => :add_to_cart, :id => product %>--Eric Gruber

#32396: "return salable items." is poor wording choice, as in the real world, returns happen when we don't like an item and take it back. Since you are describing the return value of a method, the wording is technically correct. Perhaps "provides a list of items for sale" might be better.--William Nace

#31714: The statement on pg 105

"The next time you restart your application (stopping and starting script/server), it will store its session data in the database. Why not do that now?"

suggests (to me at least) that my application will now work properly with cookies. Yet restarting the server as suggested and then trying to view a page in my browser results in an error.

Patient readers will discover the remedy in the next section, but it would be nice to include an indication that while a restart is necessary, there's still one more thing to do before the cookie stuff works properly.
--Chuck Hinson

#32664: When you write session :session_key => '_depot_session_id', I suggest you remind readers that in this context, session is a declaration, and not to be confused with the session hash mentioned just mentioned.--Michael Rogers

#31950: Call me picky, but the cart title is an h1 while we already have an h1 on the page in the header ("Pragmatic Bookshelf"). You change the cart title to a div called "cart-title" on page 119, but I think it's semantically an h2 both times.--Ray Schamp

#32326: Footnote 2 states that ":id=>product is idiomatic shorthand for :id=>product.id". It is *very* helpful that you point this out, and it would be great if there was one of those references to a section in Appendix A that would explain what language feature makes this happen (for those of us learning Ruby as we are learning Rails). Also, it might be better if this footnote appeared on page 102 (where this snippet of code first appears).--JRB

#31958: It's possible to get a rather hard to find error if you leave off the parenthesis on the else clause in add_product (@items << CartItem.new(product) ). I left them off (was trying to further the habit of not using parenthesis all the time in Ruby) and this causes the following error:

<path>/rails/depot/app/models/cart.rb:13: syntax error, unexpected tIDENTIFIER, expecting kEND

I'm guessing that the issues is that without the parenthesis the binding gets messed up. Might be worth mentioning, as it took me quite a while to figure out what I had messed up... after the fix, the server might need to be restarted and the sessions cleared too (I had to do this the first time, but oddly enough, not after trying to replicate the error).--David Kapp

#31960: I got a 'uninitialized constant Cart::CartItem)' error, which took me quite a while to figure out. It was obviously a class loading problem, but I couldn't figure out why. Turns out I had put my CartItem class in a file called CartItem.rb, rather than cart_item.rb.

This is the first place that this error might occur in the book, so I think a callout might be a good idea.--Antony Blakey

#32093: When I got to the line:

>rake db:sessions:clear

I had to hit cntrl-c first to stop the server which made me wonder if stopping the server was enough to clear the sessions. Hmmm, let me think, if session information is kept in the database then the database needs to know which session to look up so I guess the browser has the session ID stored & sends it again so restarting the server has no effect unless you run the above command. If this is not the case then some mention of what is happening might round out your lesson.

#31949: While I understand why you chose to build the cart with sessions to illustrate a point, I think it would be helpful to show a counter example and build it the right way. Then when I finish that section I see the bad things that can happen if you store lots of stuff in the session, and then how exactly to do it the right way, so I can concretely see the benefits over the bad way.--Ray Schamp

#32421: "respond_to { |format| format.js }"

I saw this line of code and had trouble understanding why "format.js" and not "add_to_cart.js". Only after I googled it I found out that it's a convention and that the name of the variable didn't matter, only the ".js" did.

Wouldn't it be better if this was explained?--Breno Gazzola

#32157: We've seen a lot of hashes where a :symbol is used as a key, but here we see attributes["style"], where the key is a string. Perhaps some explanation as to why a string is used vs. a symbol?--Alex

#33054: maybe it should be better to change the code to:
respond_to do |wants|
wants.html {redirect_to_index}
wants.js
end

So people get used to doing that instead of doing request.xhr?--pufnstuff

#33092: For those out there trying to follow along with the book but use postgresql rather than sqlite3, then the migration file that creates line_itmes needs to look like this (took me several hours to figure this out):

class CreateLineItems < ActiveRecord::Migration
def self.up
create_table :line_items do |t|
t.integer :product_id, :null => false, :options =>
"CONSTRAINT line_items_product_id_fkey FOREIGN KEY (product_id) REFERENCES products(id)"
t.integer :order_id, :null => false, :options =>
"CONSTRAINT line_items_order_id_fkey FOREIGN KEY (order_id) REFERENCES orders(id)"
t.integer :quantity, :null => false
t.decimal :total_price, :null => false, :precision => 8, :scale => 2

t.timestamps
end
end

def self.down
drop_table :line_items
end
end--Bill Colvin

#32311: The top of this page shows code :

class LineItem < ActiveRecord::Base
belongs_to :order
belongs_to :product
def self.from_cart_item(cart_item)
li = self.new
li.product = cart_item.product
li.quantity = cart_item.quantity
li.total_price = cart_item.price
li
end
end

However, at this stage the from_cart_item helper method has not been introduced. It is introduced and code supplied on page 152.--Salim Jordan

#32251: The version of _cart.html.erb shown on this page does not have Ajax for the "empty cart" button, where as most users following the example would have the Ajax version at this point. --defucius tai

#32280: "<%= button_to "Checkout", :action => :checkout %>": Put a "this line changed" triangle in front of this line, to show the reader which line they need to add for this step.--Brent Nordquist

#32312: Running the SQLite3 SQL

depot> sqlite3 -line db/development.sqlite3
SQLite version 3.4.0
Enter ".help" for instructions
sqlite> select * from orders;

Shows many fields that would not be there for those following along in the book (e.x ship_class, created_at, updated_at, customer_email, etc.)

#32230: What happens if you are on the place order screen and click empty-cart? Should a user be able to order an empty cart? Implement a re-direct to the catalog. --Rick Cook

#32470: On this page, for the first time, we see the following code:
self.salt = self.object_id.to_s + rand.to_s

With this explanation:
"Without the self., Ruby would have thought we were assigning to a local variable, and our code would have no effect."

Well, I can understand why "self.salt" instead of just "salt" but not the reason for the sudden change to "self.salt" from "@salt". I believe that an explanation is necessary, especially because the User class doesn't have a "salt=" accessor.--Breno Gazzola

#32225: Seventeenth line from bottom:
"flash.now[:notice] = ..." => An explanation of "flash.now" as well as why it is used rather than just "flash" may be helpful.--dc

#33034: When I got to this page and looked at the nice login form in figure 11.1, then compared it to the completely plain one in my application, I realized that we never made an admin layout. Maybe that should have been obvious, but I missed it. Just copying the users layout and editing it seemed to do the trick.--Shane Walker

#32346: [sorry; should have posted this here first. I posted it over in the forums section before realizing there was a separate errata system. No need to reply to the forum thread on this.]

The business about making the store layout the default layout for the application is sequenced oddly, in such a way as to generate some confusing errors that aren’t mentioned in the text.

If the instructions are followed in the order that the book lists them, then when the user visits localhost:3000/users immediately after putting the layout “store” instruction in application.rb, you get a route error (or at least, I got a route error doing this on Windows).

The text should say to remove to unnecessary products.html.erb and users.html.erb BEFORE telling the user to go look at either of those controllers.

Also-although I’m sure you know this-the illustrations for that section need to be fixed because the browser screenshots in them show the wrong URLs, which was also quite confusing for me until I remembered the note at the front of the book saying not to trust the illustrations.
--Jason Black

#32386: In the def for who_bought you've changed the respond_to from |format| 2 pages prior to |accepts| without any changes as far as I can tell. If the change was merely cosmetic, I found it to be confusing.
Thanks,
-Irene--Irene Ros

#32389: Why the switch to mysql from sqlite3? The depot application has been using sqlite3 and it doesn't quite seem consistent (or necessary) to switch without at least an explanation.

Thanks!
-Irene--Irene Ros

#31750: Generated functional tests of rails 2.0.2 aren't subclass of Test::Unit::TestCase, but ActionController::TestCase.
I suppose the book should be illustrate new with latest information, and I would like to see how we override setup and teardown.

--Hiroshi Saito

#32391: You seem to have pasted the code twice for the ApplicationController separated by a single line... it doesn't really make sense. It starts on 207 and goes to 208.
Thanks!--Irene Ros

#31848: There is no mention about the creation of depot_r/test/integration/dsl_user_stories_test.rb It would be nice if there was, even if it was only a single sentence.--Shaun McDonald

#33060: At the end of section 15.3 (String Extensions) it says: "In a Rails application, these changes can go in the file environment.rb in the config directory."
Wouldn't it be a bit cleaner to put custom inflections into "config/initializers/inflections.rb"?--Bart van der Zwet

#32387: When authoring a migration for the purpose of "creating data that’s a convenience for the developer when they play with the application", it seems that the migration should be written so that it does not affect a production database. A comment here about how to check what environment the migration is running under (or perhaps some other way to do this) might be appropriate here.--JRB

#32288: "Be warned: the only data you should load in migrations is data that you’ll also want to see in production: lookup tables, predefined users, and the like. Do not load test data into your application this way.":

This question has been on my mind ever since you had me do exactly that on pages 88-90. By creating the "add_test_data" migration for the Depot sample application, it looks like you've set the stage for my development test data to ultimately end up on the production server, right?

I think it would be good to add a footnote or other explanation on pp. 88-90 about not doing it this way for a real application, and/or showing the reader here a different method to populate data for the "development" environment that will not affect the "production" environment. (What would that method be? I'm still wondering.)--Brent Nordquist

#32293: "Be warned: the only data you should load in migrations is data that you’ll also want to see in production: lookup tables, predefined users, and the like. Do not load test data into your application this way.": Is it possible to put an if statement into up function to determine the environment (development or production) and act according? --Ozum Eldogan

#32406: The illustrations for the One-to-One and One-to-Many relationships are correct, but the readability could be tweaked.

I tend to read from left to right and I usually assume the parent comes before the child. You may want to re-arrange your illustrations on the One-to-One relationships and the One-to-Many relationships, so that the parent table is to the left of the child tables (with the arrow connector pointing to the left).

#32349: "Note that it isn’t possible to set attributes in the intermediate table using this
approach." This leaves a huge gap. So how do you set the intermediate table? Maybe refer to another page???--Ray Daly