Tuesday, 23 December 2008

Unit Testing Using Context

Jeremy McAnally has released a gem called context allowing contexts in Test::Unit tests. It's all the rage, even those crazy Rails guys are doing it (although I did check out some of their unit tests earlier and couldn't find any).

Anyway, I'm way behind on my TDD and BDD and all other testing acronyms so I decided to give it a go. These are way simple tests and I actually did it the normal Rails way initially before "converting" to the context way of doing it.

I've got converting up there in quotes because it was really, really simple and I hardly did anything which is fantastic because the easier something that's good is to do, the more people will eventually do it; think Rails conventions.

Getting Started

Get the gem:
$ gem sources -a http://gems.github.com
$ sudo gem install jeremymcanally-context
That is outlined as the install procedure, but you can't just use it as you must require that bad boy for your Rails app to use it - so put this in your test_helper.rb:
require 'context'
This allows it to be used in all tests that require the test_helper. I've also put some other helper methods (where else) in here, just simple stuff like:

def get_post(*id)
if id.nil?
Post.new
else
Post.find(id)
end
end
I also noticed that in your individual Unit Tests you don't need to explicitly put
fixtures :comment
and the like because
fixtures :all
is declared in this file allowing you to just access your fixtures like so:
comment = comments(:comment_1)
Here I have a comments.yml fixture which can be accessed via
comments
and I have an individual record
:comment_1


On to the test, finally

So here's my edited Unit test class:
require File.dirname(__FILE__) + '/../test_helper'

class CommentTest < Test::Unit::TestCase

context "A new comment" do
test "should be valid with both name and body" do
comment = Comment.new(:name => 'My User',
:email => 'myuser@example.com',
:url => 'http://www.example.com',
:body => "Here's my comment for ya!")
assert comment.valid?
assert comment.errors.empty?
end

test "should be invalid with name but no body" do
comment = Comment.new(:name => 'My User',
:email => 'myuser@example.com',
:url => 'http://www.example.com')
assert !comment.valid?
assert comment.errors.invalid?(:body)
end
end

context "An existing comment" do
test "should be valid" do
comment = comments(:comment_1)
assert comment.valid?
assert comment.errors.empty?
end
end
end
As you can see it's pretty much the same as any normal Rails Unit test, except for a couple of minor changes / improvements:
  1. The class inherits from Test::Unit::TestCase instead of ActiveSupport::TestCase
  2. We've added two context blocks in there to wrap our tests, creating a more readable and elegant test
  3. We're not
    def
    ining our test methods, we instead have test blocks, which, again, makes the whole thing read like a sentence; some would say the sign of beautiful code
  4. We can also have before and after blocks (my before block is kinda lame but yours could do whatever you want, creating multiple objects for use in your tests for example, so you don't have to create new ones for every test) much like setUp and tearDown in JUnit
On the second and third points; if you read your tests out loud they should flow like a sentence. So in the simple tests I have created we have:

"A new comment should be invalid with no body or name" - the context and the test name form a flowing sentence. "An existing comment should be valid" etc.

All of my tests start with the word should, but there is no reason they couldn't start with anything else as long as they create a sentence like structure. They could be rewritten "A new comment could be accepted as long as a user is present", it's a bit more verbose, but it still reads like a sentence and everyone looking at that test will know what it's meant to be testing.

There is still a lot to learn, like Mocks, Integration Tests, Functional Tests and even more on Unit Tests - like writing them before the code.

But my first foray was an enjoyable one and something I will definitely keep doing and I hope this is a help to even just one person and gives them a boost to create their own tests, making their software more stable and maintainable.

Thoughts, comments, corrections and/or improvements welcome.

Rebuild the locate database on Mac

Just because I always forget.
$ sudo /usr/libexec/locate.updatedb

Sunday, 9 November 2008

Merb 1.0

So Merb 1.0 was released 2 days ago during RubyConf, just
sudo gem install merb
to update.

I had to update my config/dependencies.rb file to read
merb_gems_version = "1.0"
and also
sudo gem install webrat
, which is a Ruby Acceptance Testing plugin. Here's the Git repo if you want to check it out.

Thursday, 6 November 2008

Getting the Database name in Merb and DataMapper

After a lot of searching and banging my head against my keyboard I finally found how to get the database name for my Merb application:
Macintosh:ruby-rss-feeds marktucks$ merb -i
...
>> repository(:default).adapter
=> #, @name=:default, @transactions={}>
>> repository(:default).adapter.uri
=> #
>> repository(:default).adapter.uri.path
=> "sample_development.db" # YAY!

I'm sure there must be a better way to do this. A Rails::Configuration.database_configuration equivalent? But I just cannot find it. The
Merb.config
doesn't seem to have a database property / method. You can get stuff like the host (
Merb.config[:host]
).

Any suggestions / better solutions are welcome.

Friday, 26 September 2008

iTunes Library from PC to Mac

About a couple of months ago I switched from my PC to a MacBook and have never looked back since. I don't miss anything from my PC, except my iTunes Library. And contrary to my colleague's advice it seems as though you can't just copy all that iTunes guff from your PC to your Mac.

So I found this article on how to convert my iTunes library and copy all the music files from the PC to the Mac.

One small edit to this post is that in new iTunes 8, "Import Library" is now, for some confusing reason; File > Library > Import Playlist. "Consolidate Library" is also under File > Library.

It almost worked like a charm, but apparently it couldn't find some of my music, maybe something to do with the special characters in my Library.xml?

Wednesday, 27 August 2008

Ubiquity: Firefox Command Line Awesomeness

Ubiquity - "a Mozilla Labs experiment into connecting the Web with language in an attempt to find new user interfaces that could make it possible for everyone to do common Web tasks quicker and easier."

Getting started

Install the Plugin, which will restart Firefox / Mozilla.
*Note: if you are using a Mac you'll need to install Growl to get the notifications to work. Windows works out of the box, popping up those little toaster notifications in the bottom right of your screen. No messages on Linux just yet, but I'm sure there is someone out there clever enough to get this working or know how to get it working.

Once restarted, hit CTRL+SPACE for Windows or OPTION+SPACE for Mac and you'll see Ubiquity pop up below your open tabs.

There are already quite a few commands built in (type "command-list" into Ubiquity to see them all), follow this tutorial to get up and running.

One of the sweet things about it is creating your own commands using the command editor ("command-editor" in Ubiquity). Just create your command in the editor using Javascript and the tutorial provided and without needing to restart the browser, it's ready to use.

I've been mucking about with it today and created a Ruby-Doc search command just to kick the tyres:
var noun_type_ruby_keyword = {
_name: "ruby keyword",
suggest: function( text, html ) {
var suggestions = [CmdUtils.makeSugg(text)];
return suggestions;
}
}

CmdUtils.CreateCommand({
name: "ruby-api",
takes: {keyword: noun_type_ruby_keyword},
icon: "http://www.ruby-doc.org/_img/favicon.ico",
description: "Searches Ruby-Doc for the keyword(s)",
preview: function(pblock, directObject) {
searchTerm = jQuery.trim(directObject.text);

var pTemplate = "Searches Ruby-Doc for ${query}";
var pData = {query: searchTerm}
pblock.innerHTML = CmdUtils.renderTemplate(pTemplate, pData);
var url = "http://ajax.googleapis.com/ajax/services/search/web";
var params = { v: "1.0", q: searchTerm };

jQuery.get(url, params, function(data) {
var numToDisplay = 3;
var results = data.responseData.results.splice( 0, numToDisplay );

pblock.innerHTML = CmdUtils.renderTemplate( {file:"google-search.html"},
{results:results}
);
}, "json");
},
execute: function(directObject) {
var url = "http://www.google.com/search?q=site%3Awww.ruby-doc.org+{QUERY}"
var query = directObject.text;
var urlString = url.replace("{QUERY}", query);
Utils.openUrlInBrowser(urlString);
}
});

As you can see, it's got JQuery built in (along with many other helper functions) so it's easy to do Ajax and all sorts of goodness. I'm very impressed, especially as it's currently a Prototype - looking forward to future releases and the community of commands.