Question

Here’s the code, it’s very simple:

<?php
$tab = array (
    (object)array( 'id' => 1,),
    (object)array( 'id' => 4,),
    (object)array( 'id' => 12,),
    (object)array( 'id' => 22,),
    (object)array( 'id' => 25,),
);

$tab_json = array (
    (object)array( 'id' => 1,),
    (object)array( 'id' => 4,),
    (object)array( 'id' => 12,),
    (object)array( 'id' => 22,),
    (object)array( 'id' => 25,),
    (object)array( 'id' => 2,),
);
$difference = array_udiff($tab_json, $tab, function($a, $b) {
    echo $a->id." <-> ".$b->id."\n";
    return (count(array_diff_assoc(get_object_vars($a), get_object_vars($b))))>0;
});
?>

Here’s the output:

12 <-> 4
12 <-> 1
12 <-> 22
12 <-> 25
2 <-> 12
4 <-> 25
4 <-> 1
22 <-> 4
25 <-> 1
12 <-> 4
12 <-> 1
12 <-> 22
25 <-> 12
4 <-> 22
1 <-> 4
1 <-> 22
1 <-> 4
1 <-> 1
1 <-> 25
25 <-> 4
25 <-> 1
25 <-> 12
25 <-> 25
25 <-> 4
4 <-> 1
4 <-> 12
4 <-> 25
4 <-> 22
22 <-> 1
22 <-> 12
22 <-> 25
22 <-> 12
12 <-> 1
12 <-> 12
12 <-> 2
2 <-> 12
2 <-> 25

I don’t understand how it’s computed: look at 12: it’s compared more than 10 times (whereas, from what I understand, it should be compared no more than the number of elements of the second array), and moreover it’s compared three times with 1!

Tested on:

PHP 5.3.9
PHP 5.3.2-1ubuntu4.14

Answer 1

The way array_udiff() actually produces its result is of course an implementation detail and subject to change. So while what I say might be true right now, you must not rely on it.

For instance, if you need to use an old version of PHP, you might well implement a compatibility version of the function which simply compares each element of the first array against each element of the second array, discarding any that match.

On the other hand, if you want the most performance, you would quicksort the elements of the second array, and then perform a binary search of the elements of the first array.

PHP 5.3 seems to combine a quicksort with a linear search. It does however abort the search once it’s found a larger element.