Creating tables in the database

WordPress provides tables and, yet even better, functions to store the data of your plugin. There is for example wp_postmeta to store data applying to a post, repectively wp_usermeta for user data.
To store general data of your plugin you can use the wp_options table. You don't even need to create a new row for every single piece of data but you can combine all your data into an array and use the serialize() function to store it into a single set.
(Edit: Since version 2.0.1 it's not necessary to seialize the data yourself because the update_* take care of it themselves.)
As you see, there is no need to create your own tables or to use your own SQL statements.

Always requesting "fresh" data

You do not need to pull data from external sources or aggregate data from the database every time your plugin is executed. There are good chances that the data you need does not change over a longer period of time. You can use some sophisticated WordPress functions to cache the data using the "Transient API". It provides function to temporary store data (set_transient), to retrieve it (get_transient) and some others to control storage of temporary data.

Creating everything from scratch

WordPress contains a heap of useful functions especially for database access. So before you start connecting manually to the WordPress database take a look at the $wpdb class which contains everything you need to coennect to the database, add_user_meta, add_post_meta, and other useful functions to store, alter and delete data. Take a look at them before even thinking about creating your custom SQL statement. Besides being easy to use these function assure that your plugin will work even if the database layout is changed dramatically.

After me, the deluge!

Your plugin may be the best invention since sliced bread and everyone should use it. Nevertheless be convinced that there are definitely people who test you plugin and think that there must be something better and uninstall it. So take care that your plugin does no leave any clutter in the database by deleting any trace on uninstall. On uninstall not on deactivation because a user may deactivate your plugin though he wants to continue using it, e.g. for debugging purposes, and would be happy if his options are still set upon re-activation. So be nice!

There are many other things you can do wrong ® at programming a good WordPress plugin and I hope that @nacin will create a blog about it some day.

If you hover the mouse cursor over the icon all found feeds will be displayed:

As you see not only the main feed is found but also the comment feed and possible others. With this script installed you can simply click on this icon and subscribe to feeds without frantically scanning the page.

• Download “Jasper’s Google Reader subscribe”

<select size="5">
  <option "test1" />
  <option "test2" />
  <option "test3" />
  <option "test4" />
  <option "test5" />
  <option "test6" />
</select>

What everybody including me expects is something like this:

But I actually got this:

The mystery lies in the style sheet for the admin area:

#wpcontent select {
	padding: 2px;
	height: 2em;
	font-size: 12px;
}

This forces every selectbox to display only one element. I have no idea why this is implemented. There is also a ticket in the WP tracking system discussing this behaviour since two years.

The solution to this problem is quite easy:

<select style="height:auto" size="5">

Seems easy but costed me over 1 hour of my lifetime.

The problem is solved with a few lines of code:

1
2
3
4
5
6
7
8
9

add_filter( 'the_title', 'short_title' );
function short_title( $title ) {
  $chars = 20;
  if ( strlen( $title ) > $chars ) {
    $title = substr( $title, 0, $chars );
    $title .= " …";
  }
  return $title;
}

The variable $chars has to be set to the number of characters you want to have in your title and in line 6 you have to define the text to replace the omitted part of the title (… is the typographically correct version of three points, called ellipse).

It’s a bit disadvantageous that the title could be cut in the middle of a word. If you like you can extend the function accordingly:

1
2
3
4
5
6
7
8
9
10

add_filter( 'the_title', 'short_title' );
function short_title( $title ) {
  $chars = 20;
  if ( strlen( $title ) > $chars ) {
    $title = substr( $title, 0, $chars );
    $title = substr($title, 0, strrpos( $title, ' ' ) );		    $title .= " …";
  }
  return $title;
}

If you copy the function into the file functions.php of your theme all calls of the_title will return the shortened title.

Let’s see how-to remove the author’s name from the post info.

The post info is produced by this function:

function genesis_post_info() {
  global $post;
 
  if ( is_page( $post->ID ) )
    return; // don't do post-info on pages
 
  $post_info = '[post_date] ' . __('By', 'genesis') . 
               ' [post_author_posts_link] [post_comments] [post_edit]';
 
  printf( '<div class="post-info">%s</div>', apply_filters('genesis_post_info',
          $post_info) );
}

As you see the function calls a filter for the string $post_info which contains the shortcodes for the post info. So it’s easy to define a function which removes the “__('By', 'genesis') . ' [post_author_posts_link]” code:

add_filter('genesis_post_info', 'my_post_info');
function my_post_info() {
	$post_info = '[post_date] [post_comments] [post_edit]';
  return $post_info;
}

If you put the code in the functions.php of your Genesis child theme the output looks something like this:

This is all you need if you’re your blog’s only author.

link, photo, regular, video

As you see the categories are automatically linked to their respective category pages which in most cases is the style needed. Nevertheless there may be situations where you (or your customer) does not want the categories to link anywhere. Unfortunately WordPress does not provide any straightforward method to accomplish this so we have to make a small detour.

There are two ways we can take: rebuild the output or filter the output.

Rebuilding the output

$i = 0;
$sep = ', ';
$cats = '';
foreach ( ( get_the_category() ) as $category ) {
  if (0 < $i)
    $cats .= $sep;
  $cats .= $category->cat_name;
  $i++;
}
echo $cats;

(I found the basics for this solution at SydneyFX.)
The code does what we want but it has the disadvantage that it leaves out all the candy provided by the original function. So there has to be a better solution.

Filter the standard output

WordPress is full of filters and as expected there is also a filter for the output of the_category(). So let’s us it:

function no_links($thelist) {
  return preg_replace('#<a.*?>([^<]*)</a>#i', '$1', $thelist);
}
 
add_filter( 'the_category', 'no_links' );
 
[...] The Loop [..]
 
remove_filter( 'the_category', 'no_links' );

The function no_links() contains a regular expression which removes all anchors from the string. That’s the whole trick. At the end of your code you should remove the filter again, just in case the categories are displayed on the same page with the standard style.

1) Upload the entire my-cool-plugin folder to the /wp-content/plugins/ directory
2) DO NOT change the name of the my-cool-plugin folder
[...]

What the heck? Why shouldn’t I change the name of the folder? The author didn’t actually hard code the plugin path of the plugin, did he?

load_plugin_textdomain( 'my-cool-plugin', false, 'my-cool-plugin/languages' );

Yes he did! And now he’s hoping that all users won’t change the path name. But they will, Murphy’s everywhere! And after they did so they will write you an email complaining that the plugin does not work, never telling you that they changed the pathname, of course.

To avoid such trouble isn’t too difficult because you can determine the name of your plugin using some simple functions:

$plugin_path = dirname(plugin_basename(__FILE__));
load_plugin_textdomain( 'my-cool-plugin', false, $plugin_path . '/languages' );

Each part of the code will return the following:

__FILE__ -> /path-to-wordpress/wp-content/plugins/my-cool-plugin/plugin.php
 
plugin_basename(__FILE__) -> my-cool-plugin/plugin
 
dirname(plugin_basename(__FILE__)) -> my-cool-plugin

Now the user can name the plugin directory whatever he likes, your plugin is prepared.
So if you ever want to point to a directory first make sure that there is no way to determine it by the use of some function before putting it literally in your code.

In the heading I wrote “… almost never” because you can and actually have to refer to directories below your plugin directory directly in your code.

BTW: I downloaded the above mentioned plugin from a well known theme creator’s site.

There are many tutorials telling that you should rename your admin account to something else so that an attacker does not already has got a clue what account he has to break. Unfortunately WordPress is telling everyone if you have done so or not:

This blog owner was not so cautious and hasn’t changed the admin account. Now the attacker has a first clue and only needs to test his passwords against this username.
Wouldn’t it be much better if the error message would be the same no matter if the username is right or not:

Now the attacker does not know if the username is registered and has to try all his passwords even if the name is wrong.
The above error message is created by my new plugin called “Unified Login Error Messages” (or “Ulem”). It changes the two above mentioned error messages but keeps others like “The username field is empty.” untouched.
There are some tutorials telling you how to disable the error messages totally but no response to a user’s action is one of the main failures in UI design.

Download the plugin

If you’re doing any changes to the Twenty Eleven theme you should never ever modify the files in the twentyten directory. Instead create a child theme:

• Create a directory called myTwentyTen
• Create a file style.css containing:

  /*
  	Theme Name: my2010
  	Theme URL: 
  	Description:2010 test child theme
  	Author: Latz
  	Author URI: http://elektroelch.de
  	Version: 1.0
  	Template: twentyten
   
  */
   
  @import url("../twentyten/style.css");

  (Change Author: and Author URI to your needs.)

Create a file called functions.php. In this file we’re going to save the modified code for displaying the author’s name:

<?php
 
function twentyten_posted_on() {
  printf( __( '<span class="%1$s">Posted on</span> %2$s <span class="meta-sep">', 'twentyten' ),
    'meta-prep meta-prep-author',
    sprintf( '<a href="%1$s" title="%2$s" rel="bookmark"><span class="entry-date">%3$s</span></a>',
      get_permalink(),
      esc_attr( get_the_time() ),
      get_the_date()
    )
  );
}
?>

This is the original code from Twenty Ten‘s function.php with all code removed which concerns the author. Twenty Ten has the advantage that is uses this code in all places where it displays “by author”. So there is no need to change anything else.

echo '<a href="' . $link . '" id="' . $id .'" class="' . $class . '">' . 
       $linktext . '</a>';

or

?>
<a href="<?php echo $link; ?>" id="<?php echo $id; ?>" 
   class="<?php echo $class; ?>"><?php echo $linktext; ?></a>
<?php

or

echo "<a href=\"$link\" id=\"$id\" class=\"$class\">$linktext</a>";

Apart from the personal preferences of the programmer all these snippets have one thing in common: they are hard to read. “Sure!”, you will say, “attributes in HTML have to be enclosed in double quotes and in PHP I have to use one of the above methods.” Are you sure? Do attributes really have to be enclosed in double quotes? The simple answer is: No!
Let’s take a look at the specification:

[...] By default, SGML requires that all attribute values be delimited using either double quotation marks (ASCII decimal 34) or single quotation marks (ASCII decimal 39). Single quote marks can be included within the attribute value when the value is delimited by double quote marks, and vice versa. [...]

Double or single quotation marks. That will make things much easier:

echo "<a href='$link' id='$id' class='$class'>$linktext</a>";

As you see the code is much easier to read and that simply happened because you read the specs!