PHP: exceptions vs errors?
One thing to add here is about handling exceptions and errors. For the purpose of the application developer, both errors and exceptions are "bad things" that you want to record to learn about the problems that your application has - so that your customers have a better experience in the long run.
So it makes sense to write an error handler that does the same thing as what you do for exceptions.
I usually set_error_handler to a function that takes the error and throws an exception so that whatever happens i'll just have exceptions to deal with. No more @file_get_contents just nice and neat try/catch.
In debug situations i also have an exception handler that outputs an asp.net like page. I'm posting this on the road but if requested I will post example source later.
edit:
Addition as promised, I've cut and pasted some of my code together to make a sample.
<?php
define( 'DEBUG', true );
class ErrorOrWarningException extends Exception
{
protected $_Context = null;
public function getContext()
{
return $this->_Context;
}
public function setContext( $value )
{
$this->_Context = $value;
}
public function __construct( $code, $message, $file, $line, $context )
{
parent::__construct( $message, $code );
$this->file = $file;
$this->line = $line;
$this->setContext( $context );
}
}
/**
* Inspire to write perfect code. everything is an exception, even minor warnings.
**/
function error_to_exception( $code, $message, $file, $line, $context )
{
throw new ErrorOrWarningException( $code, $message, $file, $line, $context );
}
set_error_handler( 'error_to_exception' );
function global_exception_handler( $ex )
{
ob_start();
dump_exception( $ex );
$dump = ob_get_clean();
// send email of dump to administrator?...
// if we are in debug mode we are allowed to dump exceptions to the browser.
if ( defined( 'DEBUG' ) && DEBUG == true )
{
echo $dump;
}
else // if we are in production we give our visitor a nice message without all the details.
{
echo file_get_contents( 'static/errors/fatalexception.html' );
}
exit;
}
function dump_exception( Exception $ex )
{
$file = $ex->getFile();
$line = $ex->getLine();
if ( file_exists( $file ) )
{
$lines = file( $file );
}
?><html>
<head>
<title><?= $ex->getMessage(); ?></title>
<style type="text/css">
body {
width : 800px;
margin : auto;
}
ul.code {
border : inset 1px;
}
ul.code li {
white-space: pre ;
list-style-type : none;
font-family : monospace;
}
ul.code li.line {
color : red;
}
table.trace {
width : 100%;
border-collapse : collapse;
border : solid 1px black;
}
table.thead tr {
background : rgb(240,240,240);
}
table.trace tr.odd {
background : white;
}
table.trace tr.even {
background : rgb(250,250,250);
}
table.trace td {
padding : 2px 4px 2px 4px;
}
</style>
</head>
<body>
<h1>Uncaught <?= get_class( $ex ); ?></h1>
<h2><?= $ex->getMessage(); ?></h2>
<p>
An uncaught <?= get_class( $ex ); ?> was thrown on line <?= $line; ?> of file <?= basename( $file ); ?> that prevented further execution of this request.
</p>
<h2>Where it happened:</h2>
<? if ( isset($lines) ) : ?>
<code><?= $file; ?></code>
<ul class="code">
<? for( $i = $line - 3; $i < $line + 3; $i ++ ) : ?>
<? if ( $i > 0 && $i < count( $lines ) ) : ?>
<? if ( $i == $line-1 ) : ?>
<li class="line"><?= str_replace( "\n", "", $lines[$i] ); ?></li>
<? else : ?>
<li><?= str_replace( "\n", "", $lines[$i] ); ?></li>
<? endif; ?>
<? endif; ?>
<? endfor; ?>
</ul>
<? endif; ?>
<? if ( is_array( $ex->getTrace() ) ) : ?>
<h2>Stack trace:</h2>
<table class="trace">
<thead>
<tr>
<td>File</td>
<td>Line</td>
<td>Class</td>
<td>Function</td>
<td>Arguments</td>
</tr>
</thead>
<tbody>
<? foreach ( $ex->getTrace() as $i => $trace ) : ?>
<tr class="<?= $i % 2 == 0 ? 'even' : 'odd'; ?>">
<td><?= isset($trace[ 'file' ]) ? basename($trace[ 'file' ]) : ''; ?></td>
<td><?= isset($trace[ 'line' ]) ? $trace[ 'line' ] : ''; ?></td>
<td><?= isset($trace[ 'class' ]) ? $trace[ 'class' ] : ''; ?></td>
<td><?= isset($trace[ 'function' ]) ? $trace[ 'function' ] : ''; ?></td>
<td>
<? if( isset($trace[ 'args' ]) ) : ?>
<? foreach ( $trace[ 'args' ] as $i => $arg ) : ?>
<span title="<?= var_export( $arg, true ); ?>"><?= gettype( $arg ); ?></span>
<?= $i < count( $trace['args'] ) -1 ? ',' : ''; ?>
<? endforeach; ?>
<? else : ?>
NULL
<? endif; ?>
</td>
</tr>
<? endforeach;?>
</tbody>
</table>
<? else : ?>
<pre><?= $ex->getTraceAsString(); ?></pre>
<? endif; ?>
</body>
</html><? // back in php
}
set_exception_handler( 'global_exception_handler' );
class X
{
function __construct()
{
trigger_error( 'Whoops!', E_USER_NOTICE );
}
}
$x = new X();
throw new Exception( 'Execution will never get here' );
?>
Exceptions are thrown - they are intended to be caught. Errors are generally unrecoverable. Lets say for instance - you have a block of code that will insert a row into a database. It is possible that this call fails (duplicate ID) - you will want to have a "Error" which in this case is an "Exception". When you are inserting these rows, you can do something like this
try {
$row->insert();
$inserted = true;
} catch (Exception $e) {
echo "There was an error inserting the row - ".$e->getMessage();
$inserted = false;
}
echo "Some more stuff";
Program execution will continue - because you 'caught' the exception. An exception will be treated as an error unless it is caught. It will allow you to continue program execution after it fails as well.
The answer deserves talking about the elephant in the room
Errors is the old way of handling an error condition at run-time. Typically the code would make a call to something like set_error_handler before executing some code. Following the tradition of assembly language interrupts. Here is how some BASIC code would look.
on error :divide_error
print 1/0
print "this won't print"
:divide_error
if errcode = X
print "divide by zero error"
It was hard to make sure that set_error_handler would be called with the right value. And even worse, a call could be made to a separate procedure that would change the error handler. Plus many times calls were interspersed with set_error_handler calls and handlers. It was easy for code to quickly get out of control. Exception handling came to the rescue by formalizing syntax and semantics of what good code was really doing.
try {
print 1/0;
print "this won't print";
} catch (DivideByZeroException $e) {
print "divide by zero error";
}
No separate function or risk of calling the wrong error handler. The code now is guaranteed to be in the same place. Plus we get better error messages.
PHP used to only have error handling, when many other languages already had evolved to the preferable exception handling model. Eventually the makers of PHP implemented exception handling. But likely to support old code, they kept error handling and provided a way to make error handling look like exception handling. Except that, there is no guarantee that some code may not reset the error handler which was precisely what exception handling was meant to provide.
Final answer
Errors that were coded before exception handling was implemented, are likely still errors. New errors are likely exceptions. But there is no design or logic to which are errors and which are exceptions. It's just based in what was available at the time it was coded, and the preference of the programmer coding it.