[kh99]

Originally Posted by snoopytas

The render uses 'eval' 'nuff said

Kym

OK, but I ran this modified version of the benchmark:

PHP Code:

require('./global.php');
require(DIR . '/includes/adminfunctions_template.php');

$vbulletin->options['addtemplatename'] = 0;

$vbulletin->templatecache['text_bit'] = compile_template('{vb:raw text}');
$vbulletin->templatecache['text'] = compile_template('{vb:raw textbits}');
$vbulletin->templatecache['text_foreach'] = compile_template('<vb:each from="textbits" value="text">{vb:raw text}</vb:each>');

// Benchmarking bit templates
$start = microtime(true);
$textbits ='';
for ($i = 1; $i < 10000; $i++)
{
    //$templater = vB_Template::create('text_bit');
    //$templater->register('text', 'This is Text #' . $i);
    //$textbits .= $templater->render();
    $text = 'This is Text #' . $i;
    eval($vbulletin->templatecache['text_bit']);
    $textbits .= $final_rendered;
}
$templater = vB_Template::create('text');
$templater->register('textbits', $textbits);
$result1 = trim($templater->render());
$end = microtime(true);
header('Content-type: text/plain');
echo('Time taken for 10.000 Bit Templates: ' . ($end-$start) . "\n");

// Benachmarking <vb:each>
$start = microtime(true);
$textbits = array();
for ($i = 1; $i < 10000; $i++)
{
    $textbits[] = 'This is Text #' . $i;
}
$templater = vB_Template::create('text_foreach');
$templater->register('textbits', $textbits);
$result2 = trim($templater->render());
$end = microtime(true);
echo('Time taken for 10.000 Array-Elements with <vb:each>: ' . ($end-$start) . "\n");

if ($result1 == $result2)
{
    echo('Success - Both methods yielded the same result data');
}
else
{
    echo('Huh? Smth. went wrong, the results are not identical');
    echo("\n\n----\n$result1\n----\n$result2");
} 


and I got these results:

Time taken for 10.000 Bit Templates: 0.072948932647705
Time taken for 10.000 Array-Elements with <vb:each>: 0.011199951171875
Success - Both methods yielded the same result data

vs. these for the original version:

Time taken for 10.000 Bit Templates: 1.542927980423
Time taken for 10.000 Array-Elements with <vb:each>: 0.011492967605591
Success - Both methods yielded the same result data

That's 6.5 times slower vs 134 times slower, and seems to indicate that eval() isn't the main offender.