It’s tempting to just post your want ad to an online job site and pick the low bidder. But that’s a terrible idea. A lot of my business comes from people who want me to finish a project that started out with the low bidder. “Our last developer couldn’t finish this…”—I hear that a lot. Sending work offshore can sometimes work but it requires an extremely rigorous approach, excellent and constant communication, and an understanding of cultural differences which most companies can’t commit to—that’s why many offshore projects fail.

So when people ask me how to hire a developer, I point them to a process written by Derek Sivers. From what I can tell, Derek is an entrepreneur and not a computer programmer (I’ve never met him). His advice for hiring a developer is excellent. I’ve been giving similar advice to people for a long time. It boils down to this:

1. Nothing on the internet is ever finished. You will never launch if you try to start with version 10. Reduce the scope of your idea to something that you can deploy quickly: version 1.
2. Really think about your project in detail and describe how every single screen and button is going to work. Draw it out on paper or on the computer. Write an outline.
3. Pick a small portion of your project and hire a few different developers to build it. Something they can finish in a few days. Yes, you will be paying several different developers for the same work. Choose the one you like best and give them the rest of the project.

Prior to yesterday, my home page was hand-coded outside of WordPress and WordPress was responsible solely for the content in the /blog directory. (I’d created the site a long time ago and decided to add a blog to it later.) Whenever I needed to update or rearrange my portfolio, I’d get the old HTML editor out and move a bunch of markup around. Adding a new entry involved a lot of copy and paste. So I decided to add a custom post type specifically for Projects.

The first step is to register your custom post types. I did this in my theme’s functions.php file:

add_action('init', 'watson_init');

function watson_init() {
  register_post_type('project',
    array(
      'labels' => array(
        'name' => __('Projects'),
        'singular_name' => __('Project'),
        'add_new_item' => __('Add New Project'),
        'edit_item' => __('Edit Project'),
        'new_item' => __('New Project'),
        'view_item' => __('View Project'),
        'search_items' => __('Search Projects'),
      ),
      'public' => true,
      'exclude_from_search' => true,
      'has_archive' => false,
      'menu_position' => 5,
      'supports' => array(
        'title', 'editor', 'thumbnail', 'custom-fields', 'revisions'
      )
    )
  );
}

That code adds a Projects menu which allows you to manage “Projects” in the WordPress admin just like you can manage Posts and Pages. It also excludes those posts from search results and customizes which blocks are visible in the editor. You can read up on all of the options in the Codex.

From the new Projects menu, I can add, update, trash, draft, publish, and unpublish Projects just like I can with Posts. It’s very cool.

The next step was to create a custom page template to show those projects. Here’s the entire template for my home page:

<?php
/*
Template Name: Projects
*/
?>

<?php get_header(); ?>

<div id="recentprojects">
	<h2>Recent projects</h2>

	<div id="projectlisting">
		<?php
		$projects = get_posts(
			array(
				'numberposts' => -1,
				'post_type' => 'project',
			)
		);
		global $post;
		foreach($projects as $post) {
			setup_postdata($post);
			?>
			<div class="project">
				<?php
				if (has_post_thumbnail()) {
					the_post_thumbnail(array(75, 75), array('class' => 'icon'));
				}
				?>
				<div class="details">
					<h3>
					<?php
					$url = get_post_meta(get_the_ID(), 'url', true);
					if ($url) {
						printf('<a href="%s">%s</a>', $url, get_the_title());
					} else {
						the_title();
					}
					?>
					</h3>
					<?php the_content() ?>
				</div>
			</div>
			<?php
		}
		?>
	</div>

	<?php
	if (have_posts()) {
		the_post();
		the_content();
	}
	?>
</div>

<?php get_footer(); ?>

Here are the important bits: Lines 14-19 loads all of posts in the the ‘project’ post type defined earlier. Lines 20-21 setup the loop for the posts. Line 22 allows you to use functions like the_content() within your loop just like any other WordPress template. Lines 49-54 show the content associated with the actual page using this template.

That’s basically it. Register your post types, create a template. Easy as 1, 2.

Now that I’ve switched to git, you may commence moving to the next much better thing that I haven’t heard of yet.

But, to the point…

Part of the deployment process for Caravela is to create a compressed build of the CSS using yuiccompressor and save it under a new name. That new version of the CSS file is then referenced in the site header. I do this because I want the CSS to be aggressively cached by browsers to save on bandwidth and improve page load times; but I also want new CSS to take effect without waiting for caches to expire. Generating a new name for the CSS file whenever it changes solves that problem neatly.

So my deployment script does this automatically:

• Get the latest commit ID of the main.css file
• Look for a version of the css with that commit ID: e.g. main_v12345.css
• If that file doesn’t exist, build it using yuicompressor
• Store the commit ID in a file you deploy so the site knows which CSS file to serve

The key component of this is getting a commit ID. I do it by getting the output of this shell command:

git log –abbrev-commit -1 assets/css/main.css | head -1 | cut -d\  -f2

What I end up with is a unique, 7 digit version identifier based on my latest commit on that file.

Then you just need to build a new CSS file and save it using that identifier (e.g. main_v12345) (if it hasn’t already been built for a previous deployment) and teach your header file/template/view to reference the current version of the CSS file.

This works with Javascript, too, or any other file that you are both caching and versioning in git. And it isn’t important that you use yuicompressor. Your build process could just be a simple copy. The important thing is that the file has a new name so that the updated version will be loaded and cached by the user’s web browsers on their very next visit.

<Location /index.php/admin>
  ExpiresDefault "now"
</Location>

“…a kernel enhancement to confine programs to a limited set of resources. AppArmor’s unique security model is to bind access control attributes to programs rather than to users.”

My database files are in /mnt/mysql rather than /var (Amazon server). Attempting to start MySQL I got errors that MySQL couldn’t write to the temp directory. When I changed the temp directory to /tmp I got new errors that it couldn’t access the data in /mnt/mysql. All of the file permissions were correct and I was attempting to start the daemon as root anyway. In addition, errors like this showed up in the syslog:

kernel: [ 7810.088219] type=1503 audit(1286826657.774:226):  operation=”mknod” pid=12802 parent=3974 profile=”/usr/sbin/mysqld” requested_mask=”c::” denied_mask=”c::” fsuid=106 ouid=106 name=”/mnt/tmp/ibTz0Fu7″

Solution: First thing you should do is make sure all of your file and directory permissions are correct. Next, check your AppArmor configuration. The MySQL configuration for AppArmor lives at /etc/apparmor.d/usr.sbin.mysqld. Add your database and temp directories to the AppArmor configuration. Then restart the AppArmor daemon.

If you want to use a primary key that is not an auto_increment field then you need to set the _isPkAutoIncrement flag on the MySql4 object to false. For example, I created a table that uses customer_entity_id as the primary key:

class MyCompany_MyModule_Model_MySql_Subscription
    extends Mage_Core_Model_MySql4_Abstract {
  public function _construct() {
    // Table resource, primary key
    $this->_init('mymodule/mymodule', 'customer_entity_id');
    // The primary key is not an auto_increment field
    $this->_isPkAutoIncrement = false;
  }
}

If the flag is true (the default), then the save() method will always do an update when you provide a value for the primary key and an insert otherwise. That’s just what you want when the primary key is an auto_increment. If the flag is false, then Magento checks for the existence of a row with that primary key value first. If it exists, it does an update and if it doesn’t exist then it does an insert.

1. Login to your Magento admin and enable logging: Configuration | Developer | Log Settings | Enabled = Yes
2. Add Mage::log() statements to your code. For example, in your observer callback: Mage::log(“My special observer called”)
3. Watch the log files. In Linux, the tail command is very helpful for this. From the /var/log folder: tail -f system.log exception.log

Some more tips:

• Include print_r() output in the log like this: Mage::log(“myvar = ” . print_r($myvar, true))
• Add to the exception.log using Mage::logException()
• The Mage class has some useful helper functions (/app/Mage.php)
• Get the actual class name of any object: $name = get_class($object)
• If you copy and paste this code, you’ll probably need to fix the fancy quotation marks.

Scratch

Scratch is a programming language that makes it easy to create your own interactive stories, animations, games, music, and art — and share your creations on the web.

Alice

Alice is an innovative 3D programming environment that makes it easy to create an animation for telling a story, playing an interactive game, or a video to share on the web.

Processing

Processing is an open source programming language and environment for people who want to create images, animations, and interactions.

I couldn’t find a good introductory video for this one. But the documentation and tutorial pages are top notch.