Categories
Uncategorized

Implementing the Observer Pattern in C#: Code Samples

I have been reading about data binding in windows phone 7 development and got curious about how .net implements the observer pattern. I spent some time understanding the flow and decided to write a blog post about it.

.NET platform offers IObserver and IObservable interfaces to quickly get up and running with the observer pattern. We know the theory – there is an independent subject, a observable aimed towards monitoring that subject and some observers which are very keen to receive updates whenever something interesting happens to the subject. The observers know that the observable monitors the subject, so they subscribe to the observable for notifications. The observable accepts all these subscriptions, keeps them in a safe place. Then he keeps watching and watching. When something interesting happens to the subject, he quickly gets to the list of subscribers and then notifies them one by one.

Compare this scenario to the day before Eid Ul Fitr. Tomorrow is Eid if we see the moon. Now, you have decided to go to the roof and keep waiting there until the moon pops up. Knowing this, some of your friends (who are so computer freak that they don’t even want to leave their PCs to watch the moon of Eid Ul Fitr) called you and asked you to notify them if you see the moon.

Here:
The Moon == Subject
You == Observable
Your nerd friends == Observers

Here the observable keeps watching for the subject, when the subject pops up in the sky, the observable calls the observers – that’s it, simply that’s the observer pattern.

To implement this pattern, we define a Subject object. It has a counter. Every time you view the counter, it gets increased by one. We are doing this little magic to track any changes in the subject.

Then we define the SubjectObservable object which implements the IObservable. It accepts a subject in it’s constructor. This is not mandatory but we did this to track the subject in a better fashion. It provides a Subscribe() method which is called by the observers to subscribe to the observers. The method must return an IDisposable object. We do a trick and create a custom object implementing the IDisposable interface. In the constructor we accept the list of observers recorded by the observable and the instance of the observer passed for subscription. In the Dispose() method we remove the observer from the observer list. So when the Dispose() method of the disposer is called, the observer is removed from the observer list. The observable will no longer send any notifications to the removed observer.

The SubjectObserver is pretty simple. It has OnNext(), OnCompleted() and OnError() methods. The OnNext() is called by the observable to send the current state of the subject. Besides the abstract methods, we also define a constructor to accept a unique name instance and an instance of an observable to follow. When the OnNext() is called, we print out the current counter value of the subject. We also define a Dispose() method which shall invoke the Dispose() method of the disposer, removing itself from the observer list of the observable.

Well, if all these theories seem too complex to you. Do read the codes below. I have tried to add relevant comments. If you have any feedback, please do make a comment.

Okay, so every time we read the value of the counter data, the value increases. The subject is inspected by the observable and the observable passes the subject to the observers. Each observer again prints the counter and increases the value by one. This continues. 🙂

Categories
PHP

Running PHPUnit on Git hook

I use Git and I love Git! I knew about Git hooks but never used in real life. So today I decided to get my hands dirty with Git hooks. I had a feeling it’d be awesome if I could use the “pre-commit” hook to do some routine processing before the changes are committed. I had two things in mind: 1) Unit Testing with PHPUnit or 2) Generating documentation using DocBlox. I chose Unit testing favouring simplicity.

Summary: I have a git repo for my PHP project. I have written some unit tests for the project and I want to run an automated unit test before changes are committed. If the unit tests succeed, the changes will be committed, if one of the tests fails, the commit will not proceed. This will ensure that I am always committing *working* codes to my Git repo.

Understanding Git Hooks
Git hooks are nothing but executable scripts which are run by Git on special events. If the hook script returns an exit code of 0, Git continues with whatever it was doing. For any non zero exit code, Git halts the operation. It’s that simple!

Git hooks are stored in the “.git/hooks” directory of a git repo. There are some samples already generated for us to check out! The script should be named with the hook, that is for the “pre-commit” hook the file name should also be “pre-commit”.

My Lame Unit Test
To begin with, we need at least one test case. So I wrote this really lame unit test and saved it as “Mytest.php”

Try running the script with PHPUnit like this:

Output:

Okay, so we have a valid unit test that passes. Now let’s add a Git pre-commit hook to run this unit test before commit is made.

Setting Up Git Hook
Assuming that your project is already under Git DVCS, navigate to .git/hooks and create a file named “pre-commit” with the following contents:

PS: The pre-commit script is a modified version of this nice Gist: https://gist.github.com/975252

Make the “pre-commit” file executable by issuing this command:

NB: This is important that you make the script executable otherwise it will not function as a hook.

Making a Commit
We shall create some file or make some changes and try to commit. In my case, I just added the “Mytest.php” file and tried to commit. Here’s the output:

Aha! All tests passed and the commit was made. Now let’s force the unit test to fail. Change the unit test to look like this:

And let’s commit now:

Compare the output with previous output. There is nothing similar to:

The commit was not made because the test failed. So, it worked! 😀

I am really loving what it is going to offer me! I shall be writing tests for most of my projects so I never unintentionally break something in a commit. Unit testing is just one type of validation, I believe we could use lots of others (eg. PHP Code Sniffer) 🙂

Categories
Linux PHP

Installing PHPUnit on Ubuntu

Summary: I am on Ubuntu 11.10 and have the necessity to install PHPUnit. From my past experience, I know that installing via repository installs old version + fails to run. So I decided to install it via PEAR. Here’s a walk through!

Upgrade PEAR
Damn! The shipped version of pear via the “php-pear” package is outdated. The first thing we need to do is upgrade our PEAR setup. We have to ask pear to upgrade itself!

That should upgrade our pear to the latest version available!

Setup PHPUnit
To install PHPUnit, we have to set auto discovery turned on and discover the PHPUnit PEAR channel. Then we force install PHPUnit with all dependencies. That’s simple!

If everything goes okay, we have a working setup of PHPUnit by now! Type “phpunit –version” on console and check out the output! Ooops! It didn’t work! We’re getting an error message regarding “require_once()”. The issue is obvious, the PEAR directory which contains all PEAR packages haven’t been added to the php include_path.

Fix it this way:

Vim editor window will open with the empty buffer. If you’re not comfortable with vim, just use another text editor. Type in the following directives:

Save and quit. We had to do this manually because pear doesn’t automatically set the include_path to the pear location (which is “/usr/share/php/”).

Now try on terminal:

The output should look like:

Happy Unit testing!