Federico Cargnelutti

Cloud Computing is rising fast, with its data centers growing at an unprecedented rate. However, this has come with concerns of privacy, efficiency at the expense of resilience, and environmental sustainability, because of the dependence on Cloud vendors such as Google, Amazon, and Microsoft. Community Cloud Computing makes use of the principles of Digital Ecosystems to provide a paradigm for Clouds in the community, offering an alternative architecture for the use cases of Cloud Computing. It is more technically challenging to deal with issues of distributed computing, such as latency, differential resource management, and additional security requirements. However, these are not insurmountable challenges, and with the need to retain control over our digital lives and the potential environmental consequences, it is a challenge we must pursue.

Towards Community Cloud Computing (Visit Site | Download PDF)

Java has no goto statement, to break or continue multiple-nested loop or switch constructs, Java programmers place labels on loop and switch constructs, and then break out of or continue to the block named by the label. The following example shows how to use java break statement to terminate the labeled loop:

public class BreakLabel
{
    public static void main(String[] args)
    {
        int[][] array = new int[][]{{1,2,3,4},{10,20,30,40}};
        boolean found = false;
        System.out.println("Searching 30 in two dimensional int array");

        Outer:
        for (int intOuter = 0; intOuter < array.length ; intOuter++) {
            Inner:
            for (int intInner = 0; intInner < array[intOuter].length; intInner++) {
                if (array[intOuter][intInner] == 30) {
                    found = true;
                    break Outer;
                }
            }
        }

        if (found == true) {
            System.out.println("30 found in the array");
        } else {
            System.out.println("30 not found in the array");
        }
    }
}

Use of labeled blocks in Java leads to considerable simplification in programming effort and a major reduction in maintenance.

On the other hand, the C continue statement can only continue the immediately enclosing block; to continue or exit outer blocks, programmers have traditionally either used auxiliary Boolean variables whose only purpose is to determine if the outer block is to be continued or exited; alternatively, programmers have misused the goto statement to exit out of nested blocks.

What’s interesting is that Python rejected the labeled break and continue proposal a while ago. And here’s why:

Guido van Rossum wrote:

I’m rejecting it on the basis that code so complicated to require this feature is very rare. While I’m sure there are some (rare) real cases where clarity of the code would suffer from a refactoring that makes it possible to use return, this is offset by two issues:

1. The complexity added to the language, permanently.

2. My expectation that the feature will be abused more than it will be used right, leading to a net decrease in code clarity (measured across all Python code written henceforth). Lazy programmers are everywhere, and before you know it you have an incredible mess on your hands of unintelligible code.

But what’s more interesting is that the idea of adding a goto statement was never ever mentioned.

Common sense perhaps?

Mather Corgan, president of HotPads, gave a great talk on how HotPads uses AWS to run their real estate search engine. HotPads abandoned their managed hosting in December and took the leap over to EC2 and its siblings. The presentation has a lot of detail on costs and other things to watch out for, so if you’re currently planning your “cloud” architecture, you’ll find some of this really helpful.

HotPads on AWS

This manual gathers together the key insights into API design that were discovered through many years of software development on the Qt application development framework at Trolltech (now part of Nokia). When designing and implementing a library, you should also keep other factors in mind, such as efficiency and ease of implementation, in addition to pure API considerations. And although the focus is on public APIs, there is no harm in applying the principles described here when writing application code or internal library code.

The Little Manual of API Design (PDF)

Part 1 – An Alternative to Zend_Controller: Introduction
Part 2 – An Alternative to Zend_Controller: The Router

Domain Layer

The domain layer is separated from the other layers and has no dependencies on Zf_Controller.

app/
domain/
    Model/
        User.php
        UserDao.php
lib/
    Zend/
    Zf/

Adding the domain directory to your include path allows you to load and access Model classes from anywhere within your project:

Bootstrap file

set_include_path(APPLICATION_PATH . '/../lib'
    . PATH_SEPARATOR . APPLICATION_PATH . '/../domain);

Index Controller

class IndexController extends Zf_Controller_Action
{
    public function indexAction()
    {
        // DAO pattern
        $dao = new Model_UserDao();
        ...
    }
}

The getInstance() method can be used to set or retrieve a singleton instance of a Model class:

class IndexController extends Zf_Controller_Action
{
    public function indexAction()
    {
        $dao = $this->getInstance('Model_UserDao');

        // or...
        $dao = new Model_UserDao();
        $this->setInstance($dao);

        $this->userAction();
    }

    public function userAction()
    {
        $dao = $this->getInstance('Model_UserDao');
        ...
    }
}

Below is an example of the UserDao class that contains methods to insert, update, and retrieve data:

class Model_UserDao extends Zf_Persistence_Db_Adapter
{
    public function find($id)
    {
        $db = $this->getAdapter();
        ...
    }
}

abstract class Zf_Persistence_Db_Adapter
{
    protected $_db = null;

    public function setAdapter(Zend_Db_Adapter_Abstract $db)
    {
        $this->_db = $db;
    }

    public function getAdapter()
    {
        if (null === $this->_db) {
            if (!Zend_Registry::isRegistered('Zend_Db')) {
                throw new Zf_Persistence_Exception('...');
            }
            $this->setAdapter(Zend_Registry::get('Zend_Db'));
        }

        return $this->_db;
    }
}

More Info

• Domain-Driven Design with Zend Framework
• Model Infrastructure

Part 1 – An Alternative to Zend_Controller: Introduction

Here’s the good news: Zf_Controller doesn’t have a Router. It uses a very simple mapping to determine the name of the controller and action. Optionally, you may define parameters in the URI, for example:

http://domain.com/controller/action/value1/value2

An example of how routes are matched:

URI: http://domain.com/news/tag/zend-framework

Controller

class NewsController extends Zf_Controller_Action
{
  public function tagAction()
  {
    print_r($this->_getAllParams());
  }
}

Outputs

Array
(
    [controller] => News
    [action] => tag
)

Variable Defaults

It is often helpful to be able to assign default values to parameters that aren’t passed, that way we don’t have to check if the parameter has been set or not. The $_params property is an array with keys representing variable names and default values.

URI: http://domain.com/news/tag/zend-framework

Controller

class NewsController extends Zf_Controller_Action
{
  protected $_params = array(
    'name' => null,
    'show' => 10
  );

  ...
}

Outputs

Array
(
    [controller] => News
    [action] => tag
    [name] => zend-framework
    [show] => 10
)

A parameter sent via the URI will always overwrite a default parameter. That’s the whole point of using default values.

Variable Requirements

The $_paramsMatch property is used to set variable requirements. These are defined as parts of a regular expression:

URI: http://domain.com/news/tag/zend-framework/50

Controller

class NewsController extends Zf_Controller_Action
{
  protected $_params = array(
    'name' => null,
    'show' => 10
  );

  protected $_paramsMatch = array('show' => '\d+');

  ...
}

In the example above, Zf_Controller will assign the value to the “show” variable only if the value is a number. If not, it will assign the default value, in this case “10″.

If you think that you need more flexibility than this, I suggest extending Zend_Controller_Action instead.

Part 3 – An Alternative to Zend_Controller: The Model

Zend Framework is very flexible and one of its strengths is that it allows developers to implement their own components. The Zend_Controller component, for example, is very powerful. Of course, it’s not my intention to replace it, but to offer an alternative that decreases the number of decisions a developer needs to make when developing an application.

Meet Zf_Controller. The Zf_Controller component has the following goals:

• Abstract complexity: Try to reduce the level of details so the developer can focus on a few concepts at a time.
• Emphasize Convention over Configuration: Emphasize CoC, meaning that the user only needs to specify unconventional aspects of the application.
• Maintain backwards compatibility: Allow the developer to replace Zf_Controller with Zend_Controller in case the application grows in size or complexity.
• Improve performance: Load less classes, execute less code.
• Remove circular references: Avoid circular references.
• Remove Singleton classes: Avoid implementing the Singleton pattern.
• Research: Learn more about the framework, what it does, how it works.

Project Structure

Zend_Controller allows you to use the project structure that best suits your needs. On the other hand, Zf_Controller is more rigid, it only allows you to use the standard project structure:

project/
  app/
    config/
    controllers/
      ErrorController.php
      IndexController.php
    views/
      layouts/
      scripts/
        error/
        index/
          index.phtml
  domain/
    Model/
      Example.php
  lib/
    Zend/
    Zf/

Zf_Controller classes:

Zf/
  Controller/
    Action/
      Helper/
        Layout.php
    Action.php
    Front.php

Bootstrap File

Zend_Controller:

define('APPLICATION_PATH', realpath(dirname(__FILE__) . '/../app'));
set_include_path(APPLICATION_PATH . '/../lib'
    . PATH_SEPARATOR . get_include_path());

require_once "Zend/Loader.php";
Zend_Loader::registerAutoload();

$frontController = Zend_Controller_Front::getInstance();
$router = $frontController->getRouter();
$routes = include_once APPLICATION_PATH . '/config/routes.php';
$router->addRoutes($routes);

$layout = Zend_Layout::startMvc();
$layout->setLayoutPath(APPLICATION_PATH . '/views/layouts');

$frontController->dispatch();

Zf_Controller (no Router):

define('APPLICATION_PATH', realpath(dirname(__FILE__) . '/../app'));
set_include_path(APPLICATION_PATH . '/../lib'
    . PATH_SEPARATOR . APPLICATION_PATH . '/../domain);

require_once "Zend/Loader.php";
Zend_Loader::registerAutoload();

$frontController = new Zf_Controller_Front();
$frontController->setLayoutPath(APPLICATION_PATH . '/views/layouts');
$frontController->dispatch();

Action Controller

The default action controller and the default action are named “index”:

class IndexController extends Zf_Controller_Action
{
  public function indexAction()
  {}
}

Error Controller

Zend_Controller:

class ErrorController extends Zend_Controller_Action
{
  // Action used by Zend_Controller
  public function errorAction()
  {
    $error = $this->_getParam('error_handler');
    echo $error->exception->getMessage();
  }
}

Zf_Controller:

class ErrorController extends Zf_Controller_Action
{
  // Action used by Zf_Controller
  public function indexAction($e)
  {
    echo $e->getMessage();
  }
}

Rendering a View Script

Zf_Controller does not use the ViewRenderer helper class. To make the code more readable and testable, you need to call the render() method and return a value:

class IndexController extends Zf_Controller_Action
{
  public function indexAction()
  {
    $view = $this->initView();
    $view->message = 'Hello';

    // Renders views/scripts/index/index.phtml
    return $this->render();
  }

  public function testAction()
  {
    $view = $this->initView();
    $view->message = 'Goodbye';

    // Renders views/scripts/index/index.phtml
    return $this->render('index');
  }}

index.phtml file:

<p><?php echo $this->message ?></p>

Using a Layout

By setting the path to your layouts in the Bootstrap file, you automatically enable the default layout “layout.phtml”:

$frontController = new Zf_Controller_Front();
$frontController->setLayoutPath(APPLICATION_PATH . '/views/layouts');

To disable the layout:

class IndexController extends Zf_Controller_Action
{
    protected $_isLayoutEnabled = false;
}

To use a different layout:

class IndexController extends Zf_Controller_Action
{
    protected $_layoutScript = 'main.phtml';
}

main.phtml file:

<!DOCTYPE html
PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
"http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
<?php echo $this->headTitle() ?>
<?php echo $this->headScript() ?>
<?php echo $this->headStyle() ?>
</head>
<body>
<div id="content"><?php echo $this->layout()->content ?></div>
</body>
</html>

Part 2: An Alternative to Zend_Controller: The Router

There’s no doubt that the additional complexity of implementing the default Front Controller in ZF results in a number of benefits. The most important ones are flexibility and extensibility. The Front Controller implementation takes into consideration the future growth of your application and its design reduces the level of effort required to extend it. A good example of this is the plugin architecture.

But, what if after evaluating the requirements of your system, you determine that there’s not sufficient complexity to implement the default Front Controller? What if you don’t need all that flexibility, a URL mapper, a ViewRendered plugin or an ActionStack helper. Maybe all you need is just a couple of controllers, and that’s it. It’s clear that the standard Front Controller does provide more options and meets every possible use case requirement, but at the cost of complexity and a lot of classes.

So, what to do? Do you choose a different tool for the job, or replace the default Front Controller with your own implementation? How hard can it be? Is it really that difficult to replace the most important component of the Zend Framework and maintain backwards compatibility?

You are about to find out.

In this series of posts, I’ll try to cover a few things I’ve learned about implementing my own Front Controller in Zend Framework. Most of the ideas and code are based on my previous posts:

• Zend Framework: The Cost of Flexibility is Complexity
• Zend Framework Automatic Dependency Tracking
• Zend Framework Controller: 22% Drop in Responsiveness
• Refactoring the Front Controller of the Zend Framework
• Improving the performance of Zend_Controller
• Zend Framework Architecture

Will simplicity finally meet power?

Part 1: Domain-Driven Design and MVC Architectures
Part 2: Domain-Driven Design: Data Access Strategies
Part 3: Domain-Driven Design: The Repository

Some of the Domain-driven design concepts explained in my previous posts are applied in this sample application. I’m also going to use the Zend Framework infrastructure to speed up some development tasks.

Directory Structure

app/     -> Application and UI Layers
domain/  -> Domain Layer
lib/     -> Infrastructure Layer

Application and UI Layers

app/
    controllers/
        UserController.php
    views/
        layouts/
        scripts/

Domain Layer

This layer should be well separated from the other layers and it should have few dependencies on the framework(s) you are using. I’m going to group all the classes and interfaces under the namespace “Project” and add it to my include path:

domain/
    Project/
        Dao/
            Db/
                IUser.php
                IUserProfile.php
                User.php
                UserProfile.php
        Model/
            User/
                IRepository.php
                Repository.php
            User.php
            UserProfile.php
            Users.php

Infrastructure Layer

This layer acts as a supporting library for all the other layers:

lib/
    Zend/
        ...
    Zf/
        Persistence/
            Db/
                Adapter.php
                Exception.php
                Replicated.php
        Domain/
            Collection.php
            Entity.php
            Exception.php
            Repository.php

User Entity

The User and UserProfile objects have a one-to-one relationship and form an Aggregate. An Aggregate is as a collection of related objects that have references between each other. Within an Aggregate there’s always an Aggregate Root (parent Entity), in this case User:

class Project_Model_User extends Zf_Domain_Entity
{
    /* SELECT id, name FROM user WHERE id = ? */
    private $properties = array(
        'id'      => null,
        'name'    => null
    );

    /* @var Project_Model_UserProfile */
    private $profile;

    public function __construct($array)
    {
        $this->populate($array);
    }
}

abstract class Zf_Domain_Entity
{
    private $properties = array();

    public function __call($method, $args) {}
    public function __set($key, $value) {}
    public function __get($key) {}
    public function __isset($key) {}
    public function __unset($key) {}
    public function populate($values) {}
    public function hasDependency($property) {}
    public function setDependency($property, $object) {}
    public function getDependency($property) {}
}

Usage:

$array = array('id'=>1, 'name'=>'Federico');
$user = new Project_Model_User($array);
$user->setProfile($profile);

echo $user->getId();   // Outputs 1
echo $user->getName(); // Outputs Federico

Users Collection

A collection is simply an object that groups multiple elements into a single unit.

class Project_Model_Users extends Zf_Domain_Collection
{
    public function __construct($users)
    {
        foreach ($users as $user) {
            if (!($user instanceof Project_Model_User)) {
                throw new Zf_Domain_Exception(...);
            }
            $this->append($user);
        }
    }
}

abstract class Zf_Domain_Collection implements Countable, Iterator
{
    ...
}

User DAO

The UserDAO class allows data access mechanisms to change independently of the code that uses the data:

class Project_Dao_Db_User extends Zf_Persistence_Db_Adapter
{
    public function find($id)
    {
        $db = $this->getAdapter();
        $query = $db->select();
        $query->from('user');
        $query->where('id = ?', $id);
        $result = $db->fetchRow($query);

        return $result;
    }

    public function findAll()
    {
        ...
        return $resultSet;
    }
}

User Repository

A Repository is basically a collection of Aggregate Roots. Collections are used to store, retrieve and manipulate Entities and Value objects, however, object management is beyond the scope of this post.

The UserRepository object injects dependencies on demand, making the instantiation process inexpensive. A caller method is responsible for dynamically creating the setter and getter methods. You can easily mock objects by passing a custom config array via the constructor.

class Project_Model_User_Repository extends Zf_Domain_Repository
{
    /* @var Project_Dao_Db_User */
    private $userDao;

    /* @var Project_Dao_Db_UserProfile */
    private $userProfileDao;

    /* @var array IoC Spec */
    private $inject = array(
        'userDao'        => 'Project_Dao_Db_User',
        'userProfileDao' => 'Project_Dao_Db_UserProfile'
    );

    public function getUserById($id)
    {
        $row = $this->getUserDao()->find($id);
        $user = new Project_Model_User($row);

        $row = $this->getUserProfileDao()->findByUserId($id);
        $profile = new Project_Model_Profile($row);
        $user->setProfile($profile);

        return $user;
    }

    public function getUsers()
    {
        $users = array();
        $rows = $this->getUserDao()->findAll();
        foreach ($rows as $row) {
            $users[] = new Project_Model_User($row);
        }

        return new Project_Model_Users($users);
    }
}

abstract class Zf_Domain_Repository
{
    ...
    public function __call($method, $arguments)
    {
    	$property = lcfirst(substr($method, 3));
        if (!property_exists($property)) {
            throw new Zf_Domain_Exception(...);
        }
        if (null === $this->$property
            && array_key_exists($property, $this->inject)) {
        	$this->$property = new $this->inject[$property];
        }
        return $this->$property;
    }
    ...
}

Usage:

$repo = new Project_Model_User_Repository();
$user = $repo->getUserById(1);
$profile = $user->getProfile();
$users = $repo->getUsers();

Links

If you’re interested in learning more about Domain-driven design, I recommend the following articles:

• Domain Driven Design and Development In Practice
• Domain-Driven Design in an Evolving Architecture

Part 2: Domain-Driven Design: Data Access Strategies

The Ubiquitous Language

The ubiquitous language is the foundation of Domain-driven design. The concept is simple, developers and domain experts share a common language that both understand. This language is set in business terminology, not technical terminology. This ubiquitous language allows the technical team become part of the business.

The Repository

Repositories play an important part in DDD, they speak the language of the domain and act as mediators between the domain and data mapping layers. They provide a common language to all team members by translating technical terminology into business terminology.

In a nutshell, a Repository:

• Is not a data access layer
• Provides a higher level of data manipulation
• Is persistence ignorance
• Is a collection of aggregate roots
• Offers a mechanism to manage entities

Injecting DAO’s

In DDD, you inject Repositories, not DAO’s in domain entities. In the absence of an ORM framework, the DAO handles the impedance mismatch that a relational database has with object-oriented techniques. For example, imagine you have an entity named User that needs to access the database to retrieve the User details:

class User
{
    // DAO needs to be injected
    private $_dao;
    private $_data;

    public function setDao(UserDaoInterface $dao) {}
    public function getDao() {}

    public function setData($data) {}
    public function getData()
    {
        if (null === $this->_data) {
            $data = $this->getDao()->find($id);
            $this->setData($data);
        }
        return $this->_data;
    }
}

The User DAO class will look something like this:

interface UserDaoInterface
{
    public function find($id);
}

class UserDaoDb implements UserDaoInterface
{
    // Db needs to be injected
    private $_db;

    public function setDb(DbInterface $db) {}
    public function getDb() {}

    public function find($id)
    {
        $db = $this->getDb();

        $query = $db->select();
        $query->from('user');
        $query->where('id = ?', $id);

        return $db->fetchRow($query);
    }
}

$userDao = new UserDaoDb();
$userDao->setDb($db);

$user = new User();
$user->setDao($userDao);
$data = $user->getData();

Is there an easier way to create entities and inject dependencies? The process of creating an entity is complex, because an entity always has relationship with other objects in your domain model. When creating an entity, we have to initialize its relationships as well. Therefore, it’s a good practice to delegate this task to another object.

But, what about separation of concerns? DAO’s are related to persistence, and persistence is infrastructure, not domain. The main problem with the example above is that we have lots of different concerns polluting the domain. According to DDD, an object should be distilled until nothing remains that does not relate to its meaning or support its role in interactions. And that’s exactly the problem the Repository pattern tries to solve.

Injecting Repositories

Lets create a UserRepository class to isolate the domain object from details of the UserDaoDb class:

class User
{
    // Repository needs to be injected
    private $_repository;
    private $_data;

    public function setRepository(UserRepositoryInterface $repo) {}
    public function getRepository() {}

    public function setData($data) {}
    public function getData()
    {
        if (null === $this->_data) {
            $data = $this->getRepository()->getUserById($id);
            $this->setData($data);
        }
        return $this->_data;
    }
}

It’s the responsibility of the UserRepository to work with all necessary DAO’s and provide all data access services to the domain model in the language which the domain understands.

class UserRepositoryInterface
{
    public function getUserById($id);
}

class UserRepository implements UserRepositoryInterface
{
    // DAO needs to be injected
    private $_userDao;

    public function setUserDao(UserDaoInterface $userDao) {}
    public function getUserDao() {}

    public function getUserById($id)
    {
        return $this->getUserDao()->find($id);
    }
}

$userDao = new UserDaoDb();
$userDao->setDb($db);

$userRepo = new UserRepository();
$userRepo->setUserDao($userDao);

$user = new User();
$user->setRepository($userRepo);
$data = $user->getData();

The main difference between the Repository and the DAO is that the DAO is at a lower level of abstraction and doesn’t speak the ubiquitous language of the domain.

So, what’s next?

We’ve seen how to write a persistence-ignorant domain model. Next, I’ll explain how to automate the creation and injection of dependencies.

Part 4: Zend Framework Domain-Driven Design