Tech Notes From The Trenches

how to ignore *.php requests using Apache

2010-09-01T19:33:29Z

I recently deployed a beta version of one of my projects.

A few days went by and then I started getting 404 email notifications to my inbox. It seems that someone was trying to exploit known PHP issues (eg. phpMyAdmin issues, as seen in the picture below)

Here is how I solved this issue:

Make sure you have mod_rewrite enabled for your Apache
Write a rule in your VirtualHost that looks like this:

1
2
3
4
5
6

<VirtualHost *:80>
ServerName example.com
DocumentRoot /path/to/example.com/current/public
RewriteEngine On
RewriteRule ^.*.php$ 404.html [R]
</VirtualHost>

This will redirect any *.php requests to your 404.html and save resources in your server.

getting started with acts_as_state_machine

2010-08-26T16:04:23Z

This is a note to self. Remember to define all states for an object before you start referencing it.

If not, you will get an error like this:
“You have a nil object when you didn’t expect it! The error occurred while evaluating nil.entering”

This is the solution:

1
2
3
4
5
6
7
8
9
10
11

class Squad < ActiveRecord::Base
acts_as_state_machine :initial => :in_progress

event :complete do
transitions :from => :in_progress, :to => :completed,
:guard => Proc.new {|s| !s.completed_at.blank? }
end

state :in_progress # This is the line I had forgotten to write

state :completed, :enter => :do_complete

Have fun with acts_as_state_machine!

And keep in mind that acts_as_state_machine is not concurrent. Read more here.

how to stub methods in cucumber

2010-08-05T15:55:16Z

Most of the Cucumber tests that I’ve written don’t need stubbing. But the other day I was trying to test a particular feature which integrated with an external service.

Then, stubbing a call to the external service seemed like the best way to go. Fortunately, since Cucumber 0.8.4, mocking and stubbing is now possible without hacks. See: Mocking and Stubbing with Cucumber.

Here is the gist.

Add this line to your env.rb

1	require 'spec/stubs/cucumber'

Start stubbing methods in your steps.rb files

For example:

1	ExternalService.should_receive(:send).and_return(my_response)

1	User.should_receive(:customer_add_success?).and_return(true)

Next thing I’d like to do is use stubbing to test features in the feature. For example: Test a feature involving recurring events around daylight time savings time.

SVG Javascript chart libraries

2010-07-20T20:14:30Z

The last couple of weeks I worked on a feature that required to display a set of charts. Namely, a pie chart and a bar chart. One of them had to be interactive. We didn’t want a Flash chart.

I did some research and found these libraries:

This is what I think of them.

gRaphaël

I liked it from the beginning. I thought it was simple and straightforward. I started using it. But the lack of documentation made me look for something better.

One would have to search through all the issues to find a solution. The examples weren’t extensive enough for me. And I didn’t want to scan through the source code to find out how to tweak it.

I would re-consider this library once they write the documentation. (Right now it says ‘work in progress’)

Highcharts

This library looks beautiful. But the license for it is a bit expensive. And I’m working on a commercial project. If there wasn’t a free solution, I would consider it.

Google Chart Tools

I love this library. It is very well documented. There are a lot of examples. There is even a Code Playground section!

A few things that could be better:

The customization of the look and feel. It could be more customizable.
If it didn’t generate an iframe per chart.
It could be faster.

But this library was good enough for my feature.

“Aptana Unified Builder” problem & solution

2010-07-07T17:56:26Z

I ran into a problem today with RadRails 2 and its ‘Aptana Unified Builder.’ I was trying to launch my project and the IDE kept hanging on “Invoking ‘Aptana Unified Builder’ on ‘/project’”

After updating my RadRails2 with a nightly build and trying different versions of RadRails, Aptana and even RubyMine; this is my proposed solution.

Go to Preferences > Run/Debug > Launching.
Switch ‘Wait for ongoing build to complete before launching’ to ‘Never.’
Switch off ‘Build (if required) before launching’
Press OK

That solved my problem. Now back to work!

how to work with git local branches

2010-06-25T15:43:58Z

Compared to SVN and CVS, I think the best thing about Git is Github. It’s awesome how you can see the history of your source code.

So here are the steps to work with local git branches.

1. Commit and push the latest code from your local environment

1 2	[etagwerker@gesell prode]$ git commit -m "My latest stable release" [etagwerker@gesell prode]$ git push

2. Pull the latest code from the server

1 2	[etagwerker@gesell prode]$ git pull --rebase Current branch master is up to date.

3. Create a new local branch

1 2	[etagwerker@gesell prode]$ git checkout -b second_round Switched to a new branch 'second_round'

4. Modify the code, add a feature, set of features or correct bugs, or whatever you want to do. Finally, commit all the code to your local branch

5. Switch to the master branch

1 2	[etagwerker@gesell prode]$ git checkout master Switched to branch 'master'

6. Pull the latest code from the master branch

1
2
3
4

[etagwerker@gesell prode]$ git pull origin master
From github.com:/etagwerker/prode
* branch master -> FETCH_HEAD
Already up-to-date.

7. Merge the local branch with master

1
2
3
4
5
6
7
8
9
10
11

[etagwerker@gesell prode]$ git merge second_round
Updating 3043225..7ca2025

Fast forward

app/controllers/application_controller.rb | 5 +++
app/models/forecast.rb | 4 ++
config/initializers/constants.rb | 4 ++-
.../20100625123027_add_round_number_to_games.rb | 16 ++++++++
10 files changed, 94 insertions(+), 13 deletions(-)
create mode 100644 db/migrate/20100625123027_add_round_number_to_games.rb

8. Push the code to your remote branch

1
2
3
4
5
6
7
8

[etagwerker@gesell prode]$ git push
Counting objects: 44, done.
Delta compression using up to 2 threads.
Compressing objects: 100% (25/25), done.
Writing objects: 100% (25/25), 3.43 KiB, done.
Total 25 (delta 18), reused 0 (delta 0)
To git@github.com:/etagwerker/prode.git
3043225..7ca2025 master -> master

9. And you are done!

Now you can work on local branches and merge with master whenever you want.

getting started with Cucumber and Webrat on Rails

2010-06-17T19:01:36Z

A couple of months ago I started using Cucumber in one of my projects. I think it’s a great tool for large projects. It makes it easier for developers joining the project and it gives the development team a lot of confidence when changing the code.

If you have a running set of tests, you know that you can go in, write a failing test, modify the code, add a feature and then run the tests again. If the tests fail, you broke something that was working: fix it. If you didn’t, great news: you have a new working feature!

There is a great Railscasts that explains how to get started: http://railscasts.com/episodes/155-beginning-with-cucumber

Here is a set of steps to follow.

1. Create your Rails app

rails blog

2. Add Cucumber and Webrat to your environment.rb

config.gem 'cucumber',    :lib => false,
  :version => '=0.4.3'
config.gem 'webrat',      :lib => false,
  :version => '>=0.5.3'
config.gem 'cucumber-rails',      :lib => false,
  :version => '>= 0.3.1'

3. Install your gems

cd blog
sudo rake gems:install

4. Generate your Cucumber code

cd blog
./script/generate cucumber --webrat

3. Start writing feature files

For example:

vi features/list_posts.feature

Feature: Listing all the posts in the blog
  As a user
  I want to see all the posts in the blog

  Scenario: I go to the posts list page and I see all the posts
    Given there is a post titled "Awesome post" with content "My cool content"
    When I go to the posts page
    Then I should see what is up
    And I should see "Awesome post"
    And I should see "My cool content"
    And I should not see "New post"

4. Start fixing failing tests

Cucumber will show you:

Passing steps
Skipped steps: These steps weren’t evaluated due to failing or undefined step
Failing steps: These steps don’t pass. Something isn’t working as it should
Undefined steps: These steps aren’t defined in features/step_definitions/*.rb – You should define it or make sure you spelled this step correctly.

Webrat provides you a lot of useful steps to get started. You can see them in features/step_definitions. I encourage you to go in and check it out.

5. Create features/step_definitions/general_steps.rb and put this in:

Then /^I should see what is up$/ do
  save_and_open_page
end

6. Create features/step_definitions/blog_steps.rb and put this in:

Given /^there is a post titled "([^\"]*)" with content "([^\"]*)"$/ do |title, content|
  Post.new(:title => title, :content => content).save
end

Note: After modifying the database, you will need to run rake db:migrate on your dev environment and then rake db:test:prepare to clone the changes to the cucumber database.

Have fun cuking!

please stop overdesigning!

2010-06-11T14:21:58Z

Keep it simple, stupid. Or keep it simple stupid. Or keep it simple and straightforward.

But please, do it for other programmers. Do it for yourself, for your own sanity. Stop overdesigning and ‘designing for the future.’ Stop designing for ‘what may happen in the future.’

You don’t need utterly complicated designs. You are not an oracle and you cannot tell what your application will need to do in the future. So please solve a concrete problem and move on.

95% of the time you are overdesigning your code. Solve a problem now. Refactor later. Thank you!

rails-authorization-plugin first impressions

2010-06-01T16:56:54Z

I started using the rails-authorization-plugin last week. It is a really awesome and flexible plugin. I don’t have much to say about it yet.

But there is one thing that I find confusing:

In order to set a role for a user, you can do it like this:

user.is_admin_for(group)

To check if a user has a role, you can do it like this:

user.is_admin_for?(group)

See the subtle difference? Well, I think the difference is too subtle. Is there a better way to set a role for a user? Haven’t found it yet.

Other than that, the plugin is awesome. It’s super flexible and the integration is pretty seamless. Can’t wait to learn more about it.

migrating from SQLite to PostgreSQL via Heroku

2010-05-20T15:17:19Z

I had been using SQLite in my development environment, and PostgreSQL in the beta environment for one of my projects, until today. I decided to change to PostgreSQL.

I think it’s a good (and almost essential) practice to use the same tools/libraries on development, beta and production.

This is what I did:

1. Created two PostgreSQL databases locally (app_development & app_test) with pgAdmin3

2. Created a PostgreSQL user (app_user) with ‘create database’ privileges

3. Changed my database.yml to look like this

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15

development:
adapter: postgresql
database: app_development
username: app_user
password: app_pwd
timeout: 5000
encoding: utf8

test: &TEST
adapter: postgresql
database: app_test
username: app_user
password: app_pwd
timeout: 5000
encoding: utf8

4. Ran heroku db:pull

That created all the tables in my database and pulled the data from beta.
This is a very useful way to get new development environments up and running, using Heroku.