Devguru

Rails Bad Practices #2 - using Time.now in scope

Wed, 22 Jun 2011 16:54:15 +0200

Ever wondered how to get your time-dependent scopes to act like bunch of lunatics?

do this:


class User < ActiveRecord::Base

 scope :active, where(:activated_at.gt => Time.now)

end

and your list of active users will stub every time you restart your app. Cool!

Of course, this great trick can’t be achieved with this:


class User < ActiveRecord::Base

 scope :active, lambda{ where(:activated_at.gt => Time.now) }

end

Rails Bad Practices #1 - sql injection.

Fri, 10 Jun 2011 11:15:00 +0200

Sql injection is a very nice trick, however it’s really hard to achieve with Rails. This few simple lines of code allow all of your users to execute sql whatever way they want:


#app/models/user.rb
class User < ActiveRecord::Base

  scope :very_clever_scope, lambda{|name_or_id| where("name = #{name_or_id} OR id=#{name_or_id}") 

end


#app/controllers/users_controller.rb
class UsersController < ApplicationController 

  def show
    #uber clever params[:id] backwards compability
    User.very_clever_scope(params[:id]).find(:first)
  end

end

And now you can relax and enjoy DROP ALL TABLES queries.

Recursively Setting Deep Hash Value - revisited

Tue, 31 May 2011 11:19:00 +0200

This great tip allows you to recursively set hash value - really useful thing when you have to generate complicated hash.

One gotcha that i found about this is, that after generating SuperHash instance you must remember about reseting the default value of this hash.

Consider this:

 
class SuperHash < Hash   
  def initialize     
    super { |h, k| h[k] = SuperHash.new }   
  end 
end  
> suuper = SuperHash.new #=> {} 
> suuper['mad']['sheep'] = false    
> suuper['mad']['sheep'] #=> false 
> suuper['mad']['cow'] #=> {} 
> supper['calm']['sheep'] #=> {}

Notice that every undefined key, no matter how deeply nested will return empty hash instead of nil. This was a bit misleading for me since i used SuperHash instance to check complex conditions


 > !!{} #=> true

Whatever i did, all conditions were always true :)

The solution of course w’d be to reset SuperHash default after generating full structure:

 
> suuper = SuperHash.new #=> {} 
> suuper['mad']['sheep'] = false
> suuper.default = nil #=> nil
> supper['calm']#=> nil

However, as my colleague pointed out, reseting default only on suuper wont fix the issue, since nested hashes inside still have their own defaults. What we need to do is complete rewrite of our superhash back to hash again. This can be easily achievied with this:

 
> suuper = SuperHash.new #=> {} 
> suuper['mad']['sheep'] = false #=> false
> suuper = {}.update(suuper)
> supper['calm'] #=> nil
> suuper['mad']['cow'] #=> nil
> supper['calm']['sheep'] #=> Error

And now we are golden :)

RVM readline library problems?

Tue, 31 May 2011 10:50:15 +0200

4 steps to resolve:

1. sudo apt-get install ncurses-dev
2. cd $HOME/.rvm/src/ruby-1.9.2-p180/ext/readline
3. ruby extconf.rb — —with-readline-dir=”$HOME/.rvm/usr”
4. make install

More information:
http://beginrescueend.com/packages/readline/

Pow

Sat, 23 Apr 2011 23:47:57 +0200

Since most of us use macs and most of us work on few projects at the same time we really appreciate what 37signals did. Go and check out Pow project.

humans.txt

Tue, 19 Apr 2011 22:00:33 +0200

There is an interesting initiative coming from Spain. The idea is to put humans.txt file containing information about people who have contributed to building the site. More at humanstxt.org

Never run rails 3 application on ruby 1.8.7

Tue, 19 Apr 2011 13:11:46 +0200

You must update to 1.9.2. RoR3 applications on old ruby are slow. REALLY SLOW.

Incompatible character encodings with serialized fileds in ruby 1.9.2

Tue, 19 Apr 2011 13:09:32 +0200

Watchout for serialized fields in ActiveRecord. On ruby 1.9.2 strings in such fields get saved with ASCII-8BIT encoding, not with UTF-8 as you w’d expect.

Coffeescript flame.

Thu, 14 Apr 2011 23:56:00 +0200

Since we use coffeescript at netguru for some time now, we c’d sit back and relax while entire RoR community was burning in the hot fire of this flame.

Our favourite pics from the comments:

And this one pretty much sums it up:

http://is.gd/QUlhVZ

Mongoid 2.0 and its new documentation

Tue, 12 Apr 2011 13:42:32 +0200

Mongoid 2.0 and its new documentation

Beginner railscasts.

Wed, 02 Mar 2011 10:06:00 +0100

#26 - 6 min

#47 - 9 min

#48 - 10 min

#62 - 11 min

#134 - 7 min

#140 - 4 min

#152 - 9 min

#154 - 8 min

This are the numbers of Railscasts episodes that every Rails newbie should watch (obligatory for every newcommer in our company).

Webrats http_accept method missing in Capybara.

Wed, 09 Feb 2011 15:21:20 +0100

If you are missing Webrats http_accept method after migrating from Webrat to Capybara this c’d be solution for rack based driver:

page.driver.header "Accept", "application/javascript"

Alternative hash syntax in Ruby 1.9

Thu, 06 Jan 2011 11:21:38 +0100

Alternative hash syntax in Ruby 1.9

Shortcut for finding objects in rails console

Wed, 05 Jan 2011 12:55:00 +0100

Tired of typing User.find_by_verylongattribute_name(‘something’) into rails console? This little alias made my day:


class User < ActiveRecord::Base

  def self.[](arg)
    self.find_by_userlogin(arg)
  end

end

Usage:


>> User['madsheep']
=> #

Saving embedded attachments with Mongoid and Carrierwave

Wed, 29 Dec 2010 11:57:51 +0100

While developing our latest application, we discovered that Mongoid doesn’t save embedded objects when their parent is saved. On the GitHub page of this issue there were suggested some solutions and it appears that one posted by mcasimir works very well. What happened to be the next problem is that with this solution comes the issue with saving attachments’ filenames. We’ve found a solution somewhere on GitHub and changed it a little so it works with overriding filenames :

 
  after_validation :extract_filename

  def extract_filename
    img_original_filename = img.instance_variable_get("@original_filename")
    if self.img.class.filename and img_original_filename.present?
      self.img_filename = self.img.class.filename
    elsif img_original_filename.present?
      self.img_filename = img_original_filename
    end
  end

Thought of the day:

Tue, 21 Dec 2010 17:36:10 +0100

Each time someone thinks it’s a good idea to create model without timestamps or use habtm instead of hmt - Mighty God kills little kitten.

Transfering databases from one development machine to another.

Tue, 21 Dec 2010 11:45:53 +0100

http://proxylocal.com/

http://adam.heroku.com/past/2009/2/11/taps_for_easy_database_transfers/

Those two solutions can be easily combined to transfer mysql database from one development machine to another (no more nasty dump emailing/sending/sharing etc.)

On first machine:

taps server mysql://db_user:db_pass@localhost/database_name online_user online_pass -p 5000
proxylocal 5000

Last command will pop out an address (like http://tz87.t.proxylocal.com/ or similar)

And second machine (to obtain sqlite database):

taps pull sqlite://development.sqlite3 http://online_user:online_pass@tz87.t.proxylocal.com

How cool is that?

ThinkingSphinx - delta field name

Mon, 20 Dec 2010 15:46:25 +0100

set_property :delta_column => 'another_delta_field'

Adding this property to sphinx indexes on your model allows you to use different field name then ‘delta’ for delta indexing - useful when two stages of your application share the same database.

FQL in Fgraph

Sun, 12 Dec 2010 10:52:03 +0100

Very simple way to use FQL in Fgraph (https://github.com/jugend/fgraph) with query caching.

require 'digest/sha1'
module FGraph
  class Client
    def fql(query)
      @cached_query ||= {}
      @cached_query[Digest::SHA1.hexdigest(query)] ||= HTTParty.get("https://api.facebook.com/method/fql.query", :query => {:query => query, :access_token => self.options[:access_token], :format => :json} )
    end   
  end
end

Interactive Web apps - brainstorming @ Netguru

Thu, 09 Dec 2010 13:20:14 +0100

Small brainstorming about solutions to do even more interactive Web apps.