Skip to end of metadata
Go to start of metadata

<ac:macro ac:name="info"><ac:parameter ac:name="title">Temporary Location for Coding Standards Review</ac:parameter><ac:rich-text-body>
<p>The coding standards official location can be found in the <a href="http://framework.zend.com/manual/en/coding-standard.html">Zend Framework Online Manual</a> and are copied here for review and updates before being placed back into DocBook format in Subversion. The manual also contains translated versions that are not available here.</p></ac:rich-text-body></ac:macro>

<h1>Table of Contents</h1>

<ac:macro ac:name="toc-zone"><ac:parameter ac:name="location">top</ac:parameter><ac:parameter ac:name="type">list</ac:parameter><ac:parameter ac:name="style">none</ac:parameter><ac:parameter ac:name="outline">true</ac:parameter><ac:parameter ac:name="indent">20px</ac:parameter><ac:parameter ac:name="printable">true</ac:parameter><ac:rich-text-body>

<h1>Overview</h1>

<h2>Scope</h2>

<p>This document provides the coding standards and guidelines for developers and teams working on or with the Zend Framework. The subjects covered are:</p>

<ul>
<li>PHP File Formatting</li>
<li>Naming Conventions</li>
<li>Coding Style</li>
<li>Inline Documentation</li>
<li>Errors and Exceptions</li>
</ul>

<h2>Goals</h2>

<p>Good coding standards are important in any development project, particularly when multiple developers are working on the same project. Having coding standards helps to ensure that the code is of high quality, has fewer bugs, and is easily maintained.</p>

<p>Abstract goals we strive for:</p>
<ul>
<li>extreme simplicity</li>
<li>tool friendliness, such as use of method signatures, constants, and patterns that support IDE tools and auto-completion of method, class, and constant names.</li>
</ul>

<p>When considering the goals above, each situation requires an examination of the circumstances and balancing of various trade-offs.</p>

<h1>PHP File Formatting</h1>

<h2>General</h2>

<p>For files that contain only PHP code, the closing tag ("<code>?></code>") is to be omitted. It is not required by PHP, and omitting it prevents trailing whitespace from being accidentally injected into the output.</p>

<ac:macro ac:name="note"><ac:parameter ac:name="title">Important</ac:parameter><ac:rich-text-body>
<p>Inclusion of arbitrary binary data as permitted by <code>__HALT_COMPILER()</code> is prohibited from any Zend framework PHP file or files derived from them. Use of this feature is only permitted for special installation scripts.</p></ac:rich-text-body></ac:macro>

<h2>Indentation</h2>

<p>Use an indent of 4 spaces with no tab characters. Editors should be configured to treat tabs as spaces in order to prevent injection of tab characters into the source code.</p>

<h2>Maximum Line Length</h2>

<p>The target line length is 80 characters; i.e., developers should aim keep code as close to the 80-column boundary as is practical. However, longer lines are acceptable. The maximum length of any line of PHP code is 120 characters.</p>

<h2>Line Termination</h2>

<p>Line termination is the standard way for Unix text files. Lines must end only with a linefeed (LF). Linefeeds are represented as ordinal 10, or hexadecimal 0x0A.</p>

<p>Do not use carriage returns (CR) like Macintosh computers (0x0D).</p>

<p>Do not use the carriage return/linefeed combination (CRLF) as Windows computers (0x0D, 0x0A).</p>

<p>Lines should not contain trailing spaces. In order to facilitate this convention, most editors can be configured to strip trailing spaces, such as upon a save operation.</p>

<h1>Naming Conventions</h1>

<h2>Abstractions Used in API (Class Interfaces)</h2>

<p>When creating an API for use by application developers (as opposed to Zend Framework internal developers), if application developers must identify abstractions using a compound name, separate the names using underscores, not camelCase. For example, the name used for the MySQL PDO driver is 'pdo_mysql', not 'pdoMysql'. When the developer uses a string, normalize it to lowercase. Where reasonable, add constants to support this (e.g. PDO_MYSQL).</p>

<h2>Classes</h2>

<p>The Zend Framework employs a class naming convention whereby the names of the classes directly map to the directories in which they are stored. The root level directory of the Zend Framework is the "<code>Zend/</code>" directory, under which all classes are stored hierarchically.</p>

<p>Class names may only contain alphanumeric characters. Numbers are permitted in class names but are discouraged. Underscores are only permitted in place of the path separator. For example, the filename "<code>Zend/Db/Table.php</code>" must map to the class name "<code>Zend_Db_Table</code>".</p>

<p>If a class name is comprised of more than one word, the first letter of each new word must be capitalized. Successive capitalized letters are not allowed; e.g., a class "<code>Zend_PDF</code>" is not allowed, while "<code>Zend_Pdf</code>" is acceptable.</p>

<p>Zend Framework classes that are authored by Zend or one of the participating partner companies and distributed with the Framework must always start with "<code>Zend_</code>" and must be stored under the "<code>Zend/</code>" directory hierarchy accordingly.</p>

<p>These are examples of acceptable names for classes:</p>

<ac:macro ac:name="code"><ac:plain-text-body><![CDATA[
Zend_Db

Zend_View

Zend_View_Helper
]]></ac:plain-text-body></ac:macro>

<ac:macro ac:name="note"><ac:parameter ac:name="title">Important</ac:parameter><ac:rich-text-body>
<p>Code that operates with the framework but is not part of the framework, such as code written by a framework end-user and not Zend or one of the framework's partner companies, must never start with "<code>Zend_</code>".</p></ac:rich-text-body></ac:macro>

<h2>Interfaces</h2>

<p>Interface classes must follow the same conventions as other classes (see above), but must end with "<code>_Interface</code>", such as in these examples:</p>

<ac:macro ac:name="code"><ac:plain-text-body><![CDATA[
Zend_Log_Adapter_Interface

Zend_Controller_Dispatcher_Interface
]]></ac:plain-text-body></ac:macro>

<h2>Filenames</h2>

<p>For all other files, only alphanumeric characters, underscores, and the dash character ("<code>-</code>") are permitted. Spaces are prohibited.</p>

<p>Any file that contains any PHP code must end with the extension "<code>.php</code>" (except View scripts, which end in "<code>.phtml</code>" by default). These examples show the acceptable filenames for containing the class names from the examples in the section above:</p>

<ac:macro ac:name="code"><ac:plain-text-body><![CDATA[
Zend/Db.php

Zend/Controller/Front.php

Zend/View/Helper/FormRadio.php
]]></ac:plain-text-body></ac:macro>

<p>File names must follow the mapping to class names described above.</p>

<h2>Functions and Methods</h2>

<p>Function names may only contain alphanumeric characters. Underscores are not permitted. Numbers are permitted in function names but are discouraged.</p>

<p>Function names must always start with a lowercase letter. When a function name consists of more than one word, the first letter of each new word must be capitalized. This is commonly called the "camelCaps" method.</p>

<p>Verbosity is encouraged. Function names should be as illustrative as is practical to enhance understanding.</p>

<p>These are examples of acceptable names for functions:</p>

<ac:macro ac:name="code"><ac:plain-text-body><![CDATA[
filterInput()

getElementById()

widgetFactory()
]]></ac:plain-text-body></ac:macro>

<p>For object-oriented programming, accessors for object members should always be prefixed with either "<code>get</code>" or "<code>set</code>". When using design patterns, such as the Singleton or Factory patterns, the name of the method should contain the pattern name where practical to make the pattern more readily recognizable.</p>

<p>Though function names may not contain the underscore character, class methods that are declared as <code>protected</code> or <code>private</code> must begin with a single underscore, as in the following example:</p>

<ac:macro ac:name="code"><ac:plain-text-body><![CDATA[
class Zend_Foo
{
protected function _fooBar()

Unknown macro: { // ... }

}
]]></ac:plain-text-body></ac:macro>

<p>Functions in the global scope, or "floating functions," are permitted but discouraged. It is recommended that these functions be wrapped in a class and declared static.</p>

<p>Functions or variables declared with a "static" scope in a class generally should not be "private", but protected instead. Use "final" if the function should not be extended.</p>

<h3>Optional Parameters</h3>

<p>Use "null" as the default value instead of "false", for situations like this:</p>

<p>public function foo($required, $optional = null)</p>

<p><strong>when</strong> $optional does not have or need a particular default value.</p>

<p>However, if an optional parameter is boolean, and its logical default value should be true, or false, then using true or false is acceptable.</p>

<h2>Variables</h2>

<p>Variable names may only contain alphanumeric characters. Underscores are not permitted. Numbers are permitted in variable names but are discouraged.</p>

<p>For class member variables that are declared with the <code>private</code> or <code>protected</code> construct, the first character of the variable name must be a single underscore. This is the only acceptable usage of an underscore in a variable name. Member variables declared as "public" may never start with an underscore. For example:</p>

<ac:macro ac:name="code"><ac:plain-text-body><![CDATA[
class Zend_Foo
{
protected $_bar;
}
]]></ac:plain-text-body></ac:macro>

<p>Like function names, variable names must always start with a lowercase letter and follow the "camelCaps" capitalization convention.</p>

<p>Verbosity is encouraged. Variable names should always be as verbose as practical. Terse variable names such as "<code>$i</code>" and "<code>$n</code>" are discouraged for anything other than the smallest loop contexts. If a loop contains more than 20 lines of code, variables for such indices or counters need to have more descriptive names.</p>

<h2>Constants</h2>

<p>Constants may contain both alphanumeric characters and the underscore. Numbers are permitted in constant names.</p>

<p>Constant names must always have all letters capitalized.</p>

<p>To enhance readability, words in constant names must be separated by underscore characters. For example, "<code>EMBED_SUPPRESS_EMBED_EXCEPTION</code>" is permitted but "<code>EMBED_SUPPRESSEMBEDEXCEPTION</code>" is not.</p>

<p>Constants must be defined as class members by using the "const" construct. Defining constants in the global scope with "define" is permitted but discouraged.</p>

<h2>Booleans and the NULL Value</h2>

<p>Unlike PHP's documentation, the Zend Framework uses lowercase for both boolean values and the "<code>null</code>" value.</p>

<h1>Coding Style</h1>

<h2>PHP Code Demarcation</h2>

<p>PHP code must always be delimited by the full-form, standard PHP tags (although you should see the note about <ac:link ac:anchor="General"><ac:link-body>the closing PHP tag</ac:link-body></ac:link>):</p>

<ac:macro ac:name="code"><ac:plain-text-body><![CDATA[
<?php

?>
]]></ac:plain-text-body></ac:macro>

<p>Short tags are never allowed.</p>

<h2>Strings</h2>

<h3>String Literals</h3>

<p>When a string is literal (contains no variable substitutions), the apostrophe or "single quote" must always used to demarcate the string:</p>

<ac:macro ac:name="code"><ac:plain-text-body><![CDATA[
$a = 'Example String';
]]></ac:plain-text-body></ac:macro>

<h3>String Literals Containing Apostrophes</h3>

<p>When a literal string itself contains apostrophes, it is permitted to demarcate the string with quotation marks or "double quotes". This is especially encouraged for SQL statements:</p>

<ac:macro ac:name="code"><ac:plain-text-body><![CDATA[
$sql = "SELECT `id`, `name` from `people` WHERE `name`='Fred' OR `name`='Susan'";
]]></ac:plain-text-body></ac:macro>

<p>The above syntax is preferred over escaping apostrophes.</p>

<h3>Variable Substitution</h3>

<p>Variable substitution is permitted using either of these two forms:</p>

<ac:macro ac:name="code"><ac:plain-text-body><![CDATA[
$greeting = "Hello $name, welcome back!";

$greeting = "Hello {$name}, welcome back!";
]]></ac:plain-text-body></ac:macro>

<p>For consistency, this form is not permitted:</p>

<ac:macro ac:name="code"><ac:plain-text-body><![CDATA[
$greeting = "Hello $

Unknown macro: {name}

, welcome back!";
]]></ac:plain-text-body></ac:macro>

<h3>String Concatenation</h3>

<p>Strings may be concatenated using the "<code>.</code>" operator. A space must always be added before and after the "<code>.</code>" operator to improve readability:</p>

<ac:macro ac:name="code"><ac:plain-text-body><![CDATA[
$company = 'Zend' . 'Technologies';
]]></ac:plain-text-body></ac:macro>

<p>When concatenating strings with the "<code>.</code>" operator, it is permitted to break the statement into multiple lines to improve readability. In these cases, each successive line should be padded with whitespace such that the "<code>.</code>" operator is aligned under the "<code>=</code>" operator:</p>

<ac:macro ac:name="code"><ac:plain-text-body><![CDATA[
$sql = "SELECT `id`, `name` FROM `people` "
. "WHERE `name` = 'Susan' "
. "ORDER BY `name` ASC ";
]]></ac:plain-text-body></ac:macro>

<h2>Arrays</h2>

<h3>Numerically Indexed Arrays</h3>

<p>Negative numbers are not permitted as array indices.</p>

<p>An indexed array may be started with any non-negative number, however this is discouraged and it is recommended that all arrays have a base index of <code>0</code>.</p>

<p>When declaring indexed arrays with the <code>array</code> construct, a trailing space must be added after each comma delimiter to improve readability:</p>

<ac:macro ac:name="code"><ac:plain-text-body><![CDATA[
$sampleArray = array(1, 2, 3, 'Zend', 'Studio');
]]></ac:plain-text-body></ac:macro>

<p>It is also permitted to declare multi-line indexed arrays using the <code>array</code> construct. In this case, each successive line must be padded with spaces such that beginning of each line aligns as shown below:</p>

<ac:macro ac:name="code"><ac:plain-text-body><![CDATA[
$sampleArray = array(1, 2, 3, 'Zend', 'Studio',
$a, $b, $c,
56.44, $d, 500);
]]></ac:plain-text-body></ac:macro>

<h3>Associative Arrays</h3>

<p>When declaring associative arrays with the <code>array</code> construct, it is encouraged to break the statement into multiple lines. In this case, each successive line must be padded with whitespace such that both the keys and the values are aligned:</p>

<ac:macro ac:name="code"><ac:plain-text-body><![CDATA[
$sampleArray = array('firstKey' => 'firstValue',
'secondKey' => 'secondValue');
]]></ac:plain-text-body></ac:macro>

<h2>Classes</h2>

<h3>Class Declarations</h3>

<p>Classes must be named by following the naming conventions.</p>

<p>The brace is always written on the line underneath the class name ("one true brace" form).</p>

<p>Every class must have a documentation block that conforms to the phpDocumentor standard.</p>

<p>Any code within a class must be indented the standard indent of four spaces.</p>

<p>Only one class is permitted per PHP file.</p>

<p>Placing additional code in a class file is permitted but discouraged. In these files, two blank lines must separate the class from any additional PHP code in the file.</p>

<p>This is an example of an acceptable class declaration:</p>

<ac:macro ac:name="code"><ac:plain-text-body><![CDATA[
/**

  • Class Docblock Here
    */
    class Zend_Class
    {
    // entire content of class
    // must be indented four spaces
    }
    ]]></ac:plain-text-body></ac:macro>

<h3>Class Member Variables</h3>

<p>Member variables must be named by following the variable naming conventions.</p>

<p>Any variables declared in a class must be listed at the top of the class, prior to declaring any functions.</p>

<p>The <code>var</code> construct is not permitted. Member variables always declare their visibility by using one of the <code>private</code>, <code>protected</code>, or <code>public</code> constructs. Accessing member variables directly by making them public is permitted but discouraged in favor of accessor methods having the <code>set</code> and <code>get</code> prefixes.</p>

<h2>Functions and Methods</h2>

<h3>Function and Method Declaration</h3>

<p>Functions and class methods must be named by following the naming conventions.</p>

<p>Methods must always declare their visibility by using one of the <code>private</code>, <code>protected</code>, or <code>public</code> constructs.</p>

<p>Following the more common usage in the PHP developer community, static methods should declare their visibility first:</p>
<ac:macro ac:name="code"><ac:plain-text-body><![CDATA[
public static foo()

Unknown macro: { ... }


private static bar()

protected static goo()

Unknown macro: { ... }

]]></ac:plain-text-body></ac:macro>

<p>As for classes, the opening brace for a function or method is always written on the line underneath the function or method name ("one true brace" form). There is no space between the function or method name and the opening parenthesis for the arguments.</p>

<p>This is an example of acceptable class method declarations:</p>

<ac:macro ac:name="code"><ac:plain-text-body><![CDATA[
/**

  • Class Docblock Here
    */
    class Zend_Foo
    {
    /**
  • Method Docblock Here
    */
    public function sampleMethod($a)
    Unknown macro: { // entire content of function // must be indented four spaces }

/**

  • Method Docblock Here
    */
    protected function _anotherMethod()

}
]]></ac:plain-text-body></ac:macro>

<ac:macro ac:name="info"><ac:parameter ac:name="title">Please note</ac:parameter><ac:rich-text-body>
<p>Passing function or method arguments by reference is only permitted by defining the reference in the function or method declaration, as in the following example:</p>

<ac:macro ac:name="code"><ac:plain-text-body><![CDATA[
function sampleMethod(&$a)
{}
]]></ac:plain-text-body></ac:macro>

<p>Call-time pass by-reference is prohibited.</p></ac:rich-text-body></ac:macro>

<p>The return value must not be enclosed in parentheses. This can hinder readability and can also break code if a function or method is later changed to return by reference.</p>

<ac:macro ac:name="code"><ac:plain-text-body><![CDATA[
function foo()
{
// WRONG
return($this->bar);

// RIGHT
return $this->bar;
}
]]></ac:plain-text-body></ac:macro>

<p>The use of <a href="http://php.net/manual/en/language.oop5.typehinting.php">type hinting</a> is encouraged where possible with respect to the component design. For example,</p>

<ac:macro ac:name="code"><ac:plain-text-body><![CDATA[
class Zend_Component
{
public function foo(SomeInterface $object)
{}

public function bar(array $options)
{}
}
]]></ac:plain-text-body></ac:macro>

<p>Where possible, try to keep your use of exceptions vs. type hinting consistent, and not mix both approaches at the same time in the same method for validating argument types. However, before PHP 5.2, "Failing to satisfy the type hint results in a fatal error," and might fail to satisfy other coding standards involving the use of throwing exceptions. Beginning with PHP 5.2, failing to satisfy the type hint results in an E_RECOVERABLE_ERROR, requiring developers to deal with these from within a custom error handler, instead of using a try..catch block.</p>

<h3>Function and Method Usage</h3>

<p>Function arguments are separated by a single trailing space after the comma delimiter. This is an example of an acceptable function call for a function that takes three arguments:</p>

<ac:macro ac:name="code"><ac:plain-text-body><![CDATA[
threeArguments(1, 2, 3);
]]></ac:plain-text-body></ac:macro>

<p>Call-time pass by-reference is prohibited. Arguments to be passed by reference must be defined in the function declaration.</p>

<p>For functions whose arguments permit arrays, the function call may include the "array" construct and can be split into multiple lines to improve readability. In these cases, the standards for writing arrays still apply:</p>

<ac:macro ac:name="code"><ac:plain-text-body><![CDATA[
threeArguments(array(1, 2, 3), 2, 3);

threeArguments(array(1, 2, 3, 'Zend', 'Studio',
$a, $b, $c,
56.44, $d, 500), 2, 3);
]]></ac:plain-text-body></ac:macro>

<h2>Control Statements</h2>

<h3>If / Else / Elseif</h3>

<p>Control statements based on the "<code>if</code>", "<code>else</code>", and "<code>elseif</code>" constructs must have a single space before the opening parenthesis of the conditional, and a single space between the closing parenthesis and opening brace.</p>

<p>Within the conditional statements between the parentheses, operators must be separated by spaces for readability. Inner parentheses are encouraged to improve logical grouping of larger conditionals.</p>

<p>The opening brace is written on the same line as the conditional statement. The closing brace is always written on its own line. Any content within the braces must be indented four spaces.</p>

<ac:macro ac:name="code"><ac:plain-text-body><![CDATA[
if ($a != 2) {
$a = 2;
}
]]></ac:plain-text-body></ac:macro>

<p>For "<code>if</code>" statements that include "<code>elseif</code>" or "<code>else</code>", the formatting must be as in these examples:</p>

<ac:macro ac:name="code"><ac:plain-text-body><![CDATA[
if ($a != 2) {
$a = 2;
} else {
$a = 7;
}

if ($a != 2) {
$a = 2;
} else if ($a == 3) {
$a = 4;
} else {
$a = 7;
}
]]></ac:plain-text-body></ac:macro>

<p>PHP allows for these statements to be written without braces in some circumstances. The coding standard makes no differentiation and all "<code>if</code>", "<code>elseif</code>", or "<code>else</code>" statements must use braces.</p>

<p>Use of the "<code>elseif</code>" construct is permitted but highly discouraged in favor of the "<code>else if</code>" combination.</p>

<h3>Switch</h3>

<p>Control statements written with the "<code>switch</code>" construct must have a single space before the opening parenthesis of the conditional statement, and also a single space between the closing parenthesis and the opening brace.</p>

<p>All content within the "<code>switch</code>" statement must be indented four spaces. Content under each "<code>case</code>" statement must be indented an additional four spaces.</p>

<ac:macro ac:name="code"><ac:plain-text-body><![CDATA[
switch ($numPeople) {
case 1:
break;

case 2:
break;

default:
break;
}
]]></ac:plain-text-body></ac:macro>

<p>The construct "<code>default</code>" may never be omitted from a "<code>switch</code>" statement.</p>

<ac:macro ac:name="info"><ac:parameter ac:name="title">Please note</ac:parameter><ac:rich-text-body>
<p>It is sometimes useful to write a "<code>case</code>" statement which falls through to the next case by not including a "<code>break</code>" or "<code>return</code>". To distinguish these cases from bugs, such "<code>case</code>" statements must contain the comment "<code>// break intentionally omitted</code>".</p></ac:rich-text-body></ac:macro>

<h2>Inline Documentation</h2>

<h3>Documentation Format</h3>

<p>All documentation blocks ("docblocks") must be compatible with the phpDocumentor format. Describing the phpDocumentor format is beyond the scope of this document. For more information, visit <a href="http://phpdoc.org">http://phpdoc.org</a>.</p>

<p>All source code file written for the Zend Framework or that operates with the framework must contain a "file-level" docblock at the top of each file and a "class-level" docblock immediately above each class. Below are examples of such docblocks.</p>

<p>To avoid losing track of <code>@todo}}s in the source code, either use only an issue in our issue tracker, or include the issue identifier (e.g., ZF-123) in the {{@todo</code>. This allows {{@todo}}s to be tracked and monitored the same as any other issue. It also makes {{@todo}}s more visible to the community, and helps find volunteers.</p>

<p>The sharp, '#', character should not be used to start comments.</p>

<h3>Files</h3>

<p>Every file that contains PHP code must have a header block at the top of the file that contains these phpDocumentor tags at a minimum:</p>

<ac:macro ac:name="code"><ac:plain-text-body><![CDATA[
/**

<h3>Classes</h3>

<p>Every class must have a docblock that contains these phpDocumentor tags at a minimum:</p>

<ac:macro ac:name="code"><ac:plain-text-body><![CDATA[
/**

<h3>Functions</h3>

<p>Every function, including object methods, must have a docblock that contains at a minimum:</p>

<ul>
<li>A description of the function</li>
<li>All of the arguments</li>
<li>All of the possible return values</li>
<li>If a function/method may throw an exception, use "<code>@throws</code>"</li>
</ul>

<ac:macro ac:name="info"><ac:parameter ac:name="title">Please note</ac:parameter><ac:rich-text-body>
<p>It is not necessary to use the "<code>@access</code>" tag because the access level is already known from the "<code>public</code>", "<code>private</code>", or "<code>protected</code>" construct used to declare the function.</p></ac:rich-text-body></ac:macro>

<ac:macro ac:name="code"><ac:plain-text-body><![CDATA[
/**

  • Does something interesting
    *
  • @param Place $where Where something interesting takes place
  • @param integer $repeat How many times something interesting should happen
  • @throws Some_Exception_Class If something interesting cannot happen
  • @return Status
    */
    public function doSomethingInteresting(Place $where, $repeat = 1)
    Unknown macro: { // implementation...}]]></ac}

class Zend_Component_Exception extends Zend_Exception
{}

class Zend_Component_SpecificException extends Zend_Component_Exception
{}
]]></ac:plain-text-body></ac:macro>

<p>It is considered best practice within framework component code that exceptions are instantiated through the traditional <code>new</code> constructor method. </p>

<ac:macro ac:name="code"><ac:plain-text-body><![CDATA[
require_once 'Zend_Component_SpecificException.php';

class Zend_Component
{
public function foo($condition)
{
if ($condition)

Unknown macro: { throw new Zend_Component_SpecificException( 'Some meaningful exception message'); }

}
}
]]></ac:plain-text-body></ac:macro>

<p>Some concern was raised about scripts that incur overhead by loading exception classes that are by definition only used in exceptional cases. When an application's performance requirements are such that this overhead is an issue, one should use either of two solutions:</p>

<ul>
<li>Load the exception class in a traditional manner, and run the application in an environment that uses a PHP bytecode cache. A bytecode cache reduces the overhead of loading and parsing PHP classes that have been used in the environment earlier.</li>
</ul>

<ul>
<li>Lazy-load the exception class inside the code block where the exception is thrown. For example:
<ac:macro ac:name="code"><ac:plain-text-body><![CDATA[
if ($condition) {
require_once 'Zend_Component_SpecificException.php';

throw new Zend_Component_SpecificException(
'Some meaningful exception message');
}
]]></ac:plain-text-body></ac:macro></li>
</ul>

<p>Reasonable care should be taken to avoid throwing exceptions except when genuinely appropriate. In general, if a Zend Framework component is asked to perform a duty that it cannot perform in a certain situation (e.g., illegal input, cannot read requested file), then throwing an exception is a sensible course of action. Conversely, if a component is able to perform its requested duty, despite some variance from expected input, then the component should continue with its work, rather than throw an exception.</p>

<h3>Exception best practices</h3>

<ul>
<li>Use specific derived exceptions in both <code>throw</code> and <code>catch</code>. See the following two items:</li>
</ul>

<ul>
<li>Avoid throwing the <code>Exception</code> base class, or other exception superclass. The more specific the exception, the better it communicates to the user what happened.</li>
</ul>

<ul>
<li>Avoid catching the <code>Exception</code> base class, or other exception superclass. If a <code>try</code> block might encounter more than one type of exception, write a separate <code>catch</code> block for each specific exception, not one <code>catch</code> block for an exception superclass.</li>
</ul>

<ul>
<li>Some classes may require you to write more than one derived exception class. Write as many exception classes as needed, to distinguish between different types of situations. For example, "<em>invalid argument value</em>" is different from, "<em>you don't have a needed privilege</em>." Create different exceptions to identify different cases.</li>
</ul>

<ul>
<li>Don't put important diagnostic information only in the text of the exception method. Create methods and members in derived exception classes as needed, to provide information to the <code>catch</code> block. Create an exception constructor method that takes appropriate arguments, and populate the members of the class with those arguments.</li>
</ul>

<ul>
<li>Don't silently suppress exceptions and allow execution to continue in an erroneous state. If you catch an exception, either correct the condition or throw a new exception.</li>
</ul>

<ul>
<li>Keep implementation-specific exceptions isolated to the appropriate layer of your application. For instance, don't propagate <code>SQLException</code> out of the data layer code and into business layer code.</li>
</ul>

<ul>
<li>Don't use exceptions as a mechanism of flow control, or to return valid return values from a function.</li>
</ul>

<ul>
<li>Clean up resources such as database connections or network connections. PHP does not support a <code>finally</code> block as some programming languages do, so either clean up in the <code>catch</code> blocks, or else design flow control outside the <code>catch</code> block to perform cleanup, and let execution continue after the <code>catch</code>.</li>
</ul>

<ul>
<li>Use documentation from other languages for other best practices regarding using exceptions. Many of the principles are applicable, regardless of the language.</li>
</ul>

<h1>PHP Extensions</h1>

<p>Zend Framework components may use an extension only if that extension is enabled by default in the PHP.net binary distribution supported by the current release of Zend Framework. For instance, if Zend Framework 0.6.0 is intended to work with PHP 5.1.4 through 5.2.0, then the extension must be available in the default binary distribution for those versions of PHP.</p>

<p>Here is information about which extensions are enabled by default in different binary distributions of PHP on Windows:</p>

<table><tbody>
<tr>
<th><p>Extension</p></th>
<th><p>PHP 5.1.4</p></th>
<th><p>PHP 5.1.5</p></th>
<th><p>PHP 5.1.6</p></th>
<th><p>PHP 5.2.0</p></th>
<th><p>Zend Core</p></th>
</tr>
<tr>
<td><p>bcmath </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> </p></td>
</tr>
<tr>
<td><p>calendar </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> </p></td>
</tr>
<tr>
<td><p>com_dotnet</p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
</tr>
<tr>
<td><p>ctype </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
</tr>
<tr>
<td><p>date </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
</tr>
<tr>
<td><p>dom </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
</tr>
<tr>
<td><p>filter </p></td>
<td><p> </p></td>
<td><p> </p></td>
<td><p> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
</tr>
<tr>
<td><p>ftp </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> </p></td>
</tr>
<tr>
<td><p>hash </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
</tr>
<tr>
<td><p>iconv </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
</tr>
<tr>
<td><p>json </p></td>
<td><p> </p></td>
<td><p> </p></td>
<td><p> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
</tr>
<tr>
<td><p>libxml </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
</tr>
<tr>
<td><p>mysql </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> </p></td>
<td><p> </p></td>
</tr>
<tr>
<td><p>odbc </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> </p></td>
</tr>
<tr>
<td><p>openssl </p></td>
<td><p> </p></td>
<td><p> </p></td>
<td><p> </p></td>
<td><p> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
</tr>
<tr>
<td><p>pcre </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
</tr>
<tr>
<td><p>PDO </p></td>
<td><p> </p></td>
<td><p> </p></td>
<td><p> </p></td>
<td><p> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
</tr>
<tr>
<td><p>Reflection</p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
</tr>
<tr>
<td><p>session </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
</tr>
<tr>
<td><p>SimpleXML </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
</tr>
<tr>
<td><p>SPL </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
</tr>
<tr>
<td><p>standard </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
</tr>
<tr>
<td><p>tokenizer </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> </p></td>
</tr>
<tr>
<td><p>wddx </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
</tr>
<tr>
<td><p>xml </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
</tr>
<tr>
<td><p>xmlreader </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
</tr>
<tr>
<td><p>xmlwriter </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
</tr>
<tr>
<td><p>Zend Accelerator</p></td>
<td><p> </p></td>
<td><p> </p></td>
<td><p> </p></td>
<td><p> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
</tr>
<tr>
<td><p>Zend Core </p></td>
<td><p> </p></td>
<td><p> </p></td>
<td><p> </p></td>
<td><p> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
</tr>
<tr>
<td><p>Zend Debugger</p></td>
<td><p> </p></td>
<td><p> </p></td>
<td><p> </p></td>
<td><p> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
</tr>
<tr>
<td><p>Zend Optimizer</p></td>
<td><p><ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
</tr>
<tr>
<td><p>zlib </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
<td><p> <ac:emoticon ac:name="tick" /> </p></td>
</tr>
</tbody></table>

<p>The use of PHP's iconv functions are permitted and encouraged, when needed and appropriate. For example: <code>$cleanedUTF8 = iconv("UTF-8", "UTF-8//IGNORE", $badUTF8);</code></p>
</ac:rich-text-body></ac:macro>

Labels:
coding coding Delete
standard standard Delete
Enter labels to add to this page:
Please wait 
Looking for a label? Just start typing.
  1. Sep 28, 2006

    <p>Switch:</p>

    <p> "The construct "default" may never be omitted from a "switch" statement."</p>

    <p>It makes me uncomfortable to think that the equivalent of 'if(true);' is <strong>compulsory</strong> in the Zend Framework. A comment such as the following would be much better:</p>

    <p> // default: intentionally omitted</p>

    1. Oct 10, 2006

      <p>The cases that can never occur are the ones that will most certainly occur. Many debugging hours can be saved by handling such cases. If the switch is supposed to handle all values that could possibly be present, then what's the harm in adding the following:</p>

      <p> default: throw new CannotPossiblyHappenException(value);</p>

      1. Oct 10, 2006

        <p>Yes, please do add 'default' cases to catch errors, but writing</p>

        <p> // default: intentionally omitted</p>

        <p>is a little nicer for the parser than</p>

        <p> default:<br />
        break;</p>

        <p>and IMHO it's a little more obvious what's going on.</p>

        1. Oct 25, 2006

          <p>I agree with both Bill and Peter. If we do not require the style in Bill's comment, then we should at least require the style in Peter's comment. The current practice is ... odd.</p>

  2. Nov 02, 2006

    <p>Functions and Methods :</p>

    <p>maybe there most be limits for how many lines of code can exist in a method's body . if the number of lines in a method is more than the defined limit, the programmer should refactore this method and break it into smaller methods. </p>

    1. Nov 03, 2006

      <p>This should not be limited through codeing standard.</p>

      <p>It's clear that everyone of us should see to produce small and fast code. This is nothing which can be defined within a coding standard.</p>

      <p>I'll give you an example:<br />
      You define to have methods maximum 100 lines. This seems to be big and enough for standard methods within a class.</p>

      <p>But within my Date class I have to handle all PHP date values. So I have a switch statement which selects between 62 standard php values. So the function (Zend_Date->get()) is about 321 lines long including several comments, whitelines and so on...</p>

      <p>So where do you want to limit ???<br />
      100 ? I'll brake this limit without the possibility of refactoring.<br />
      400 ? You will get supersized classes when everybody just refactores when he reaches 400 lines.</p>

      <p>I think it's better to let the refactoring be done by the programmer without the use of a limit.<br />
      If someone sees the possibility for refactoring he should contact the author or write a jira issue.<br />
      This seems more naturaly for me.</p>

  3. Nov 30, 2006

    <p>What do you say about</p>

    <p>// vim: set sw=4 et :</p>

    <p>at the end of file? Is it acceptable? We have different coding standards in our company and I already have my editor associated to php extension with this setting.</p>

    1. Jan 15, 2007

      <ac:macro ac:name="code"><ac:plain-text-body><Unable to render embedded object: File ([CDATA[<) not found.--
      vim:se ts=4 sw=4 et:
      -->
      ]]></ac:plain-text-body></ac:macro>

      <p>Looks good to me, but many use IDE. I'm not sure if we can get enough support to add this to the CS above. Anyone else?</p>

    2. May 01, 2007

      <p>I strongly disagree with that because I think an editor should adapt to code and not the contrary.</p>

      <p>And what about the others editors ? Should the code include lines of configurations for dozens of editors ?</p>

  4. Nov 30, 2006

    <p>I'd like to suggest the Coding Standard be amended on file naming.</p>

    <p>Accelerator software for PHP such as Eaccelerator allows you to limit the files it compiles and caches by a pattern match on the filename.</p>

    <p>I would like to pattern match for class and interface files only, so that these will be cached while all other ".php" files may be excluded. The idea being that non class files may contain a lot of non-code PHP you don't want to be filling up your accelarator's memory cache with. It might also be desirable to differentiate such files for other reasons as well.</p>

    <p>So, I'd like to suggest all non class/interface files ending in '.php' must also be all lowercase in order to provide this pattern matching.</p>

  5. Jan 11, 2007

    <p>It seems fairly silly to me to be required to repeat the copyright, license, and perhaps even version information that's already at the top of the file again at the class. What's the point of duplicating it? It wouldn't make any sense if the file were under one license, and the class were under a different license. What kind of legal mess would you have then?</p>

    1. Apr 30, 2007

      <p>I totally concur with this. I think the comments should be normalised to eliminate logically redundant bloat. I offers no value to duplicate tags for files and classes when there is one class per file and the matadata is congruent. It means both have to be kept in sync. More overhead. There is no value in this redundancy.</p>

  6. Jan 22, 2007

    <p>In studying the changelog for PHP 5.2, I see that type hinting errors are now "PHP Catchable fatal error". Apparently, I'm not the only one irked by PHP 5.1.x's "fatal error" behavior when using type hinting: <a class="external-link" href="http://www.derickrethans.nl/erecoverableerror.php">http://www.derickrethans.nl/erecoverableerror.php</a><br />
    However, dealing with E_RECOVERABLE_ERROR from within an error handler still strikes me as inconsistent to the ZF's general approach of using catchable exceptions.</p>

    <p>Should all methods expecting instance objects of a known class always use type hinting, and always result in a fatal error, if the received argument is not of the expected class? Are there ever times when we might want to throw an exception instead?</p>

    <p>If we don't use exceptions, then the callers must be extra careful when using dynamically selected arguments (e.g. arguments selected from an array), instead of using a simple try/catch block.</p>

    <p>Since type hinting doesn't work for all data types, mixing and matching type hinting with argument validation and exception throwing would yield code blending two different ways of dealing with the same underlying issues.</p>

    <p>I'm not opposed to type hinting, but simply seeking a consistent, clear, unambiguous coding standard.</p>

    <p>Cheers,<br />
    Gavin </p>

  7. Mar 03, 2007

    <p>Methods returning a boolean property, like 'active' should start with a "is",as "isActive()".</p>

    <p>This is the same logic as for get and set,but in this case, isActive() is more obviously returning a boolean value compared to getActive()</p>

    <p>cheers<br />
    vincent</p>

  8. Apr 30, 2007

    <p>Probably not much use entering this here this late in the game, but I prefer if the identifiers are prefixed with their datatype. The exact prefixed used needn't be limited, it just has to be sensible to improve readability. For instance<br />
    VARIABLES:<br />
    $intFoo; // is an integer<br />
    $resFoo; // is a resource<br />
    $ndFoo; // is a node (for whatever context, ie. DOM node)<br />
    $elFoo; // an element<br />
    $arrFoo; // is an array<br />
    $mxdFoo; // is mixed. could be any type<br />
    $xmlFoo; // is a string containing a "well formed" XML packet<br />
    $objFoo;<br />
    $strFoo;<br />
    $tblFoo; // database results table (array of associative arrays)<br />
    $pdfFoo; // a PDF object<br />
    ...you make them up if you like, as long as it improves readability...<br />
    FUNCTIONS:<br />
    function ProcessFoo(); // returns nothing (void)<br />
    function mxdGetFoo; // returns mixed. could be any type<br />
    For obvious reasons, all classes begin with Capitals since no constructors return anything.<br />
    Obviously camelCase is the go.<br />
    I believe the benefits are obvious, so I won't elaborate.</p>

    1. Apr 30, 2007

      <p>This way of naming functions complicate refactoring since you may have to rename a method after modifying its return value.</p>

    2. May 02, 2007

      <p>Thanks for the suggestion, but we will not adopt this in the ZF coding standards.</p>

      <p>You may use conventions of your choice in your application code, but this makes no sense for the ZF to adopt it as a recommendation.</p>

  9. Apr 30, 2007

    <p>Some comments regarding this coding standards on associative arrays : Why not use this layout ?</p>
    <ac:macro ac:name="code"><ac:plain-text-body><![CDATA[
    $sampleArray = array( 'firstKey' => 'firstValue', 'secondKey' => 'secondValue',);
    ]]></ac:plain-text-body></ac:macro>
    <p>This is a little more verbose but the modified array items are easily spotted by a diff command and nicely merged by Subversion.</p>

    1. May 02, 2007

      <p>What I see in your sample layout is that the whole array is declared on one line of text.</p>

      <p>Was there a problem with the formatting of your sample layout?</p>

      <p>Putting the whole array on one line would make it <em>less</em> nice for diff. The whole line would show as changed if any array element were different. So it makes it harder to tell precisely what changed.</p>

  10. May 30, 2007

    <p>Regarding PHP extensions - for Zend Core it would also be nice to distinguish between extensions that are off by default, but can be enabled by Core and those that are not included in Core and thus can not be easily used by the Core user.</p>

  11. Jun 02, 2007

    <p>The rules on file naming are being broken in ZF's Reference Manual and the ZF Tutorial.</p>

    <p>On this page here it states:<br />
    'Any file that contains any PHP code must end with the extension ".php".'</p>

    <p>In the Reference manual in section, "7.1.2.1. Create your filesystem layout" is see:</p>

    <p> views/<br />
    scripts/<br />
    index/<br />
    index.phtml</p>

    <p>I think this needs resolving. Please also see the earlier note I posted above, as it also relates to this one when configuring accelerator software for PHP.</p>

  12. Sep 24, 2007

    <p>A section for how to handle type-casting should be written I think.</p>

    <p>Both array($value) and (array)$value are used for type-casting in Zend Framework currently - and they give quite different when you feed it with false or null making it sometimes possible to specify an attribute as null and have it treated as a zero-length array and making it sometimes result in an array containing one row with a null-value.</p>

    <p>Perhaps it would be nice to have such type casting behave the same everywhere in the framework.</p>

  13. Nov 29, 2007

    <p>I think that i was wrong that post comment in ZW coding Standard page. That is why i am posting here too, because i think it will be more correct... so...</p>

    <p>pardon me, but what operators coding standarts do you use?<br />
    we didn't find any information and in Zend FW codes there are now single style.</p>

    <p>What do you think following updates? </p>

    <p>B.4.8 Operators<br />
    All monadic(unary) operations (like "!") must be close to their operand, and all binary (like "&&") operations must be separated from both of operands by single space.</p>

    <p>if (!$b)<br />
    if ($a && $b)<br />
    if ($a && !$b)</p>

    1. Nov 29, 2007

      <p>Code such as:</p>

      <ac:macro ac:name="code"><ac:plain-text-body><![CDATA[
      if (! $object instanceof Zend_My_Interface) {
      ...
      ]]></ac:plain-text-body></ac:macro>

      <p>should be acceptable, too, under whatever guidelines we provide. Perhaps something like:</p>

      <blockquote><p>Binary operators (e.g., "<code>&&</code>") should be separated from their operands with white space, and unary operators (e.g., "<code>!</code>") may be separated from their operands with white space.</p></blockquote>

      <p>would be closer?</p>

      1. Nov 29, 2007

        <p>I agree with you that such code shuld be aceptable, but it looks like "instanseof" is exceptional situation of binary operator which priority higher than "!".It look s like  </p>
        <ac:macro ac:name="code"><ac:plain-text-body><![CDATA[
        ! $i ++ - unacceptable, but !$i++
        ! (int) $a - unacceptable, but !(int)$a
        $a && ! $b - unacceptable, but $a && !$a
        ]]></ac:plain-text-body></ac:macro>
        <p>So, would be closer following:</p>
        <blockquote>
        <p> Binary operators (e.g., "<code>&&</code>") should be separated from their operands with white space, and unary operators (e.g., "<code>!</code>", (int)) should not, they are placed close to their operand.</p>

        <p><em>NOTE:</em> Exept situation when you use "instantof" in your expressions in that way "!" may be separated from their operands with white space, for example</p>
        <ac:macro ac:name="code"><ac:plain-text-body><![CDATA[
        if (! $object instanceof Zend_My_Interface) { ...
        ]]></ac:plain-text-body></ac:macro></blockquote>
        <p>what do you think?  </p>

        1. Nov 30, 2007

          <p>Well, I also like to put a space between typecasting operators and their operands as in the following:</p>

          <ac:macro ac:name="code"><ac:plain-text-body><![CDATA[
          return (boolean) $result;
          ]]></ac:plain-text-body></ac:macro>

          <p>Is there something really wrong with this? I think that this is a stylistic choice that we should not restrict. What we should avoid is illegible code (e.g., for lack of whitespace), but we should not likely provide an exhaustive and strict set of rules governing what boils down to personal coding styles.</p>

          1. Nov 30, 2007

            <p>of course there are no any really wrong things but, you wrote:</p>

            <blockquote>
            <p>Strings may be concatenated using the "<code>.</code>" operator. A space must always be added before and after the "<code>.</code>" operator to improve readability:  </p></blockquote>
            <p>That's why i thought that you are not going to leave stylistic choise and want to provide strict set of rules. I was wrong.<br />
            I either case. thnx for answer.</p>

  14. Nov 29, 2007

    <p>And one more question about @version<br />
    As about</p>
    <blockquote>
    <p>* @version SVN: $Id:$</p></blockquote>

    <p>It's clear. we use svn:keywords and absolutely happy with svn replacing Id to full information.</p>

    <p>But, how it is better to use following construction:</p>
    <blockquote>
    <p>* @version Release: @ package_version @</p></blockquote>

    <p>thnx</p>

  15. Dec 22, 2007

    <p>Perhaps a coding standard validator might be a good idea at some point?</p>

  16. Jan 23, 2008

    <p>there is a validator in PEAR <a class="external-link" href="http://pear.php.net/package/PHP_CodeSniffer">http://pear.php.net/package/PHP_CodeSniffer</a><br />
    it may check Zend Coding Style.</p>

    <p>But of course it's much better to have possibility to check the code just in ZendStudio full file or in selected block.</p>

  17. Feb 27, 2008

    <p>There is an error in this document in the filenames section.</p>

    <p>By default view scripts are required to use the .phtml extension, yet this document states that "Any file that contains any PHP code must end with the extension ".php"." Either the code needs to be changed, or this document does.</p>

    1. Feb 28, 2008

      <p>Fixed, but these coding standards were originally intended for Zend Framework contributors (not using View scripts), not as much for Zend Framework customers (who use View scripts). Thanks! <ac:emoticon ac:name="smile" /></p>

      1. Feb 29, 2008

        <p>Darby, in trying to understand the difference in scope between this document and the "Zend Framework Coding Standards" document, it was my conclusion that the ZF Coding Standards document covered contributions to the ZF, and this document covers code which uses the ZF. After comparing the scope statements of these two documents, that was the most logical conclusion that I could surmise.</p>

        <p>If that was incorrect, then what is the difference in scope/intent of these two separate documents? Thanks!</p>

        1. Mar 01, 2008

          <p>I agree. It would be great to make this clearer.</p>

          <p>I would go further. I am personally happy to fall in with the tightest coding standards even though my code may not become part of ZF. It would still allow me to harmonise should I wish if it is clear in the first place and I think that is valuable in the long run. I do hope ZF gets a chance for some serious review/cleanup on coding standards before version 2 (see my earlier notes).</p>

          <p>Surely it is much harder to rectify these matters later on? If the criticism of PHP itself on Wikipedia is anything to go by, it really should be lesson learnt for ZF (now what was it haystack or needle first...? <ac:emoticon ac:name="smile" /></p>