Ruby On Rails

Ruby Version Manager

I’ve been meaning to start using ruby1.9 but always stuck at maintaining codes in production that depends on Ruby1.8.6. Fortunately, there’s this thing called Ruby Version Manager where you easily change between different rails version. I have nothing but praise for the guys who made life easier for us programmers to do our job better & faster with RVM.

The only thing is you have to install your gems for each ruby version. It does make sense because there are quite a few gems that depends on different built. Installing gems are not that hard imho. So no issue there.

Now I can change between different ruby version as easy as this:

$ rvm 1.9.1 # For ruby 1.9.1

$ rvm 1.8.7

What’s excellent about this is the ruby version is sandbox in just that one terminal you’re using. It doesn’t interferes with your already *original ruby installation. * rvm calls is your ruby system. To revert back to your original ruby installation just do

$ rvm system

Or if you want to set a default ruby throughout your system. You can simply do so like this:

$ rvm 1.9.1 –default

If you haven’t tried it before, go try it. Don’t worry, it won’t mess up anything on your current ruby system

Ruby Version Manager – http://rvm.beginrescueend.com/

Authlogic is Awesome

Have been tinkering around with authlogic. The most I love about is its out-of-the-box functionality. I used to wrestle with restful_authentication to get certain things that I need. Not in anyway that I’m bashing it but seeing how easy authlogic I couldn’t believe why I didn’t use it sooner.

I like it so much that I even made a rails template base on authlogic. This rails template has the usual basic signup/login/logout and I threw in some cucumber steps & stories in it. Here it is in flesh & bytes http://github.com/fadhlirahim/myrails-template/blob/master/template_authlogic.rb

My little contribution to a rails plugin (sort of)

Getting your bug fix/code pulled into the main repo really gets you excited & satisfied.

I never thought it could give me this much sense of enjoyment. Check it out.

http://github.com/dmix/weekly_builder/tree/master

Includes & Eager Loading

I’m figuring out something in my rails app.  I’m figuring out which one’s the worst:

Eager loading 100 rows of data you don’t actually need at the moment

or

10 queries giving 10 result set at once?

Here’s the code to explain it.

1
2
3
@event = Event.find(params[:event_id])
 
@event_attendees = @event.attendees.paginate(:page => params[:page], :per_page => 10)

In the view, for each attendees, it’s calling a purchase record. Since I’m using will_paginate and setting 10 records per page, for every 10 record of attendees, there’s 10 queries for a purchase record. Here’s the following SQL that will be generated:

1
2
3
4
5
6
7
8
9
10
11
12
SELECT * FROM `attendees` WHERE (`attendees`.event_id = 4)
 
SELECT * FROM `purchases` WHERE (purchasable_type = 'Attendee' AND purchasable_id = 1) LIMIT 1
SELECT * FROM `purchases` WHERE (purchasable_type = 'Attendee' AND purchasable_id = 2) LIMIT 1
SELECT * FROM `purchases` WHERE (purchasable_type = 'Attendee' AND purchasable_id = 3) LIMIT 1
SELECT * FROM `purchases` WHERE (purchasable_type = 'Attendee' AND purchasable_id = 4) LIMIT 1
SELECT * FROM `purchases` WHERE (purchasable_type = 'Attendee' AND purchasable_id = 5) LIMIT 1
SELECT * FROM `purchases` WHERE (purchasable_type = 'Attendee' AND purchasable_id = 6) LIMIT 1
SELECT * FROM `purchases` WHERE (purchasable_type = 'Attendee' AND purchasable_id = 7) LIMIT 1
SELECT * FROM `purchases` WHERE (purchasable_type = 'Attendee' AND purchasable_id = 8) LIMIT 1
SELECT * FROM `purchases` WHERE (purchasable_type = 'Attendee' AND purchasable_id = 9) LIMIT 1
SELECT * FROM `purchases` WHERE (purchasable_type = 'Attendee' AND purchasable_id = 10) LIMIT 1

Clearly you can that see the above code creates a n + 1 problem.

So I tried to modified it using eager loading like the following:

@event_attendees = @event.attendees.paginate(:include => :purchase, :page => params[:page], :per_page => 10)

Here’s the following SQL query generated:

1
2
3
SELECT * FROM `attendees` WHERE (`attendees`.event_id = 4)
SELECT `purchases`.* FROM `purchases` WHERE (`purchases`.`purchasable_id` IN (1,2,3,4..100) AND `purchases`.`purchasable_type` = 'Attendee')
 # query is shorten for brevity

You can see that it’s using 1 query that fetches 100 rows of data. At first I thought that it will limit the eager loading with my will_paginate per page settings. But clearly for every page that I’m viewing, it’s calling that query hence returning 100 rows of data for each page I’m viewing.

At the moment I’m using the query without eager loading. Simply because I’m concern the app is fetching 100 rows that I don’t need at once. I have no idea how to limit an eager loading association base on will _paginate per_page. Or probably I could eager load once and keep in cache. Or does rails does that automatically?

Will investigate though. Oh well, back to work!

Routes Gotcha

Something peculiar happened when I was doing something trivial.

If you accidently named your restful route the following

resource :roles instead of resources :roles

When you point your app to /roles it won’t go to your index controller action but it will instead call your show action.

Rebuilding Ferret Index

I had trouble reindexing the ferret index using the rake command (rake ferret:rebuild). I got this non-descriptive error regarding failing to rebuild it (Even in my development mode). So after googling for it, all one has to do is so simple.

Go into your rails script/console and just type

Model.rebuild_index

And you’re done.

Ensuring your join model uniqueness

For example:

id |category_id | inventory_id

1   384            1            #first entry
2   384            2            #this would be ok.
3   384            1            #this would not be ok

To ensure that a category_id doesn’t have any inventory_id duplicate:

[sourcecode language="ruby"]

class CategoryProduct < ActiveRecord::Base
  belongs_to :category
  belongs_to :inventory

  validates_uniqueness_of :category_id, :scope => :inventory_id
end

[/sourcecode]

Index your tables people.

A good read on table index.

http://www.railway.at/articles/2008/04/24/database-agnostic-database-ignorant

Sometimes, it is easy to forget when someone or something else handles it for you. For instance, ActiveRecords. Rails does not  create foreign keys for you. It only creates the primary key. It does not impose constraint on the database level. The association is handled on the application level. So always remember to create an index for your model associations. It could save your database server from working too hard.

Note to self

Do not name your database column with the name ‘no’.

You’ll get this error when you try to run your test

ActiveRecord::StatementInvalid: Mysql::Error: Unknown column ‘false’ in ‘field list’

Updating Ruby on Ubuntu 8.04

Pheww.. all test pass after the latest ruby package updates from the ubuntu team.

I was a little worried reading the comments from the official Rails site. From the comments, a lot of people are experiencing problems after using the patch provided by the official Ruby team.

However, today I received an update notification on my Hardy that a package is available with the latest Ruby security updates. So I installed it and no problems accounted so far while testing my Rails2.1 app. Fortunately that the project I’m working on has a good test coverage. (I made sure of it too!)