Description:
------------
As the documentation of strtr() points out, strtr "... will be the most efficient when all the keys have the same size".

Using keys of very different lengths results in poor performance, even on very small inputs.

If the str_repeat() for "m" in the test script is adjusted to 20000 the resulting runtime increases to 45 seconds for strtr() while str_replace() does not increase notably.

There are cases where the replacement array is built dynamically, so there might be little control over the keylengths. It's easy to expand the example such that strtr() takes several hours compared to just a few seconds using str_replace().

Test script:
---------------
<?php

$text = str_repeat( 'm', 2000 );

$long_from_a = str_repeat( 'a', 1 );
$long_from_x = str_repeat( 'x', 1500 );

$replacements = array(
  $long_from_a => 'b',
  $long_from_x => 'y'
);

$start = microtime( true );
$result_1 = strtr( $text, $replacements );
echo "strtr: " . number_format( microtime( true ) - $start, 4 ) . "\n";

$start = microtime( true );
$result_2 = str_replace( array_keys( $replacements ), array_values( $replacements ), $text );
echo "str_replace: " . number_format( microtime( true ) - $start, 4 ) . "\n";

echo $result_1 === $result_2 ? "results match!\n": "no match!\n";

Expected result:
----------------
strtr: 0.0001
str_replace: 0.0001
results match!

Actual result:
--------------
strtr: 2.4203
str_replace: 0.0001
results match!