But it was true, and as a Merb user I felt let down at first. All that sermonizing about the (implied) evils of Rails, and ‘we’ sell out ?! I thought, this is dodgy - why stick with the Rails label, when presumably a majority of the code is going to be merged in from Merb ?

But the Merb core team, in whom we trust, has been addressing the Merb community’s concerns, and it is becoming clear that this is a major reaffirmation of the Merb project.

My primary interest in Rails 3.0 is to get what we have in Merb 1.0: “good ruby code“, modularity, freedom of choice, and good support (availability of the core team to answer questions).

To close, the most reassuring words I have heard so far came from Yehuda who said:

Do you think for a second I would have been involved in an endeavor that represented the opposite of everything I’ve been talking about for a year now? Hell’s no!

Re-read DHH’s post. We’re all on the same page!

So, having got over the initial shock, I look forward to a cleaner Rails 3.0 based on the Merb ideals, and a guaranteed public API. Good luck to the Merb/Rails core team !

• Started working with merb + activerecord for a major project - no need to abandon ruby for django now … unless I need django-admin. My primary reason for choosing merb was its loosely coupled architecture (read: freedom of choice). I also choose to stick with AR until DM is stable enough for production use.
• Started working with git, which has been a good experience so far.
• Abandoned Gentoo Linux, though I still have a soft spot for it. What I don’t have is the time or energy to manage it. Thank you, Gentoo, for a proper linux education, but now I just need a Desktop with a low maintenance overhead. Installing Ubuntu 8.04 on my desktop and laptop was a sobering reminder of how far Desktop Linux has come.
• Reading - various stuff. One standout book that I recommend is Black Swan by Nassim Nicholas Taleb., an accessible rant about randomness and uncertainty in the real world.

My other excuse for being inactive is that I’ve been working on 2 static sites and it is sooo boring What little free time I had, I spent writing a wordpress theme for my chess blog !

I’ll profile the rails app, soon … I promise !

Efficiency

Since Ruby scans double-quoted strings for variables and escape sequences, it’s more efficient to use single-quotes for raw strings:

@sentence = 'Jackdaws love my big sphinx of quartz'
redirect_to :index => 'action'

Pretty Please, with sugar on top …

Split your strings across multiple lines. No backslashes required (sorry, couldn’t resist that !).

sql_query = "SELECT [column]
                 FROM [table]
                 WHERE [condition]
-- sql comment goes here"

or

# Q = double-quotes, embed variables 
# q = single-quotes
sql_query = %Q{SELECT @columnname
                 FROM @tablename
                 WHERE @conditions}
# You may use [ < or (as delimiters instead of {

ASCII code

To obtain the ASCII code for a character, prefix it with a question-mark:

puts ?B
# => 66

Convert it back to a string with:

puts 66.chr
# => "B"

Concatenation

Use the << operator to concatenate strings. Where += or + create a new copy, << appends to an existing string. Hence, it is more efficient:

existing << " end"

Instead of:

existing += " end"

(source)

Index a string with a regexp

You can test a string like this:

email = "spamandbakedbeans@skit.org"
if email[/@/]
  puts "I found the first @"
end

And this replaces the first match

string = "jackdaws love my big sphinx of quartz"
string[/a/] = 'A'
# => jAckdaws love my big sphinx of quartz

This replaces the second match

string = "jackdaws love my big sphinx of quartz"
string[/(j).+(z)/, 2] = "Z"
# => jackdaws love my big sphinx of quartZ

Remember, the regexp matches the entire string

Convert hex to integer (sort of)

Given a string representation of a hexadecimal, String.hex returns the corresponding integer:

puts "0x7b".hex # you can omit 0x if you wish
# => 123

Multiply a string

puts "<br />" * 3
# =&gt; "<br /><br /><br />"

require File.expand_path(File.dirname(__FILE__) + "/../../config/environment")
 
namespace :todo do
  desc 'List TODOs in all .rb files under app/'
  task(:list) do
      FileList["app/**/*.rb"].egrep(/TODO/)
  end
 
  desc 'Edit all TODOs in VIM' # or your favorite editor
  task(:edit) do
      # jump to the first TODO in the first file
      cmd = 'vim +/TODO/' 
 
      filelist = []
      FileList["app/**/*.rb"].egrep(/TODO/) {|fn,cnt,line| filelist << fn}
 
      # will fork a new process and exit, if you're using gvim
      system("#{cmd} #{filelist.sort.join(' ')}") 
  end
end

Put the above code in a .task file under lib/tasks/, and invoke them with rake todo:list, and rake todo:edit respectively.

Sample Output of rake todo:list:

app/models/xxxx_transaction.rb:13: # TODO: rewrite this ...
app/controllers/membership_controller.rb:98:  # TODO: raise an exception

The Dark Con

But for a plugin that makes authentication so easy, testing really sucks. To boot, a goldberg-enabled application is currently very unintuitive to deploy (more on that in a separate post).

Once goldberg is installed, your functional/integration tests will fail because goldberg intercepts all url requests and redirects you to /goldberg/auth/login. Fret not, here’s what to do.

Loading Goldberg’s data into the test environment

Goldberg provides two handy rake tasks:

• rake goldberg:dump_bootstrap which dumps your goldberg specific data to yaml files under vendor/plugins/goldberg/db/
• rake goldberg:load_bootstrap loads the data, from the yaml files that dump_bootstrap created, into your current database environment.

Dig into the second task and you’ll see that it uses a GoldbergMigration.load_bootstrap method.

Add this to both your functional and integration tests:

def setup
  GoldbergMigration.load_bootstrap
end

This loads the goldberg-specific data into your test environment. (of course, you need to dump it at least once using rake goldberg:dump_bootstrap !)

Integration Tests

In your integration tests, add the above code to your setup method, and use the below code in your test methods.

def test_index
 
    get "/goldberg/auth/login"
    assert_response :success
    post "/goldberg/auth/login", :login => {:name => "yourusername", :password => "yourpass"}
    follow_redirect!
    assert_response :success
 
    # do your testing here
    get "yourcontroller/index"
    assert_response :success
    assert_template "yourcontroller/index"
end

Functional Tests

Put the following into your setup method and they’ll run fine.

def setup
    get 'goldberg/auth/login' # otherwise @request.session will be empty
    @ouruser = Goldberg::User.find(:first, :conditions => 'name="yourusername"')
    Goldberg::AuthController.set_user(@request.session, @ouruser.id)
end

The downside to this is that calling GoldbergMigration.load_bootstrap for every test method slows down your tests, as it reads the .yml files from disk everytime. Awful, I know. But I can’t find a better way.

Don’t forget

Before you run your tests, don’t forget to create the proper goldberg tables in your test database: run ‘rake RAILS_ENV=test goldberg:plugin_migrations‘. Also, I am assuming you have used goldberg’s admin interface to create a user to test with - you have, haven’t you ?

Live Free and DRY Hard

For my project, I created two helper methods in test_helper.rb called goldberg_setup and goldberg_migrate. The first method has the single line (GoldbergMigration.load_bootstrap) and the second one has the code that the functional tests require.

For those who know not

The goldberg plugin, currently at version 0.2.1, is under active development, and the authors are active at the Goldberg community forum.

Thanks to unclecheese whose travails preceded mine.

Be sure to customize it to your requirements.

I wrote a rake task to import data from a csv file (output from a legacy system) into my development database for testing. The catch was that I wanted to extend my model class by adding a custom method only in the rake file.

require File.expand_path(File.dirname(__FILE__) + "/../../config/environment")
 
class Customer < ActiveRecord::Base
  def my_custom_method
  end
end

But here, for some reason, rake tries to define a _new_ activerecord model, not redefine the existing model I’ve written. So, what I tried was to precede the class definition with:

c = Customer.new # which really goes to waste

and it worked ! Rake now recognizes that I’m trying to add a method to my existing rails model !

However, I did uncover the annoyance of being unable to minimize it by clicking on the taskbar ! You have to either use the shortcut CTRL+M or Windows+D. Worse, when you maximize it, the window is ‘restored‘ instead of being maximized. I hope they fix this in the final release.

I would like:

• To view RSS item headings in a menu as in firefox. Safari opens an admittedly nifty Feed Viewer with all the feed items. But I’d really like to see the headlines at a glance.
• To use CTRL+TAB to cycle through the tabs instead of the given shortcut, which is CTRL+{ or CTRL+}. On my keyboard, this entails pressing three keys: CTRL + SHIFT + [
• A lighter theme ? The current one is too dark and gray. An option to switch to the Macintosh theme would be nice.
• To be able to click on an item in the auto-completion menu for the address bar. You have to press ENTER in Safari.
• The status bar to be visible, by default, upon installation.

Otherwise, I like Safari.

• Compared to IE, Safari is swift (to the naked eye).
• Font smoothing is configurable.
• There is an amnesiac mode called Private Browsing where, like TorPark, Safari does not record anything in the History or Downloads and form entries are not saved in AutoFill.
• History items are accessible date-wise from a recursive menu (History -> ‘Wednesday, June 13th 2007′ -> Item …. ), which, in my opinion, is far easier than opening a sidebar and navigating a tree menu.
• I haven’t noticed any website rendering incorrectly, so far.

I need Firebug at work, so I’ll probably stay with Firefox. I’m on Linux at home, so I don’t have the option of Safari there.

That said, I’ll definitely give Safari’s final release another chance.

Models created in a test fixture don’t have their created_at attributes auto-updated. They need to be specified like this:

created_at: <%=Time.now.to_s(:db)%>

Thanks to toretore on #rubyonrails (irc.freenode.net) for this tip.