Unusual PHP array iteration behavior

Question

I'm experiencing unusual behavior in PHP while iterating over arrays, and would love either an explanation or a confirmation that this is a bug.

For starters, here's the PHP version:

PHP 7.2.3 (cli) (built: Mar  8 2018 10:30:06) ( NTS )
Copyright (c) 1997-2018 The PHP Group
Zend Engine v3.2.0, Copyright (c) 1998-2018 Zend Technologies
    with Zend OPcache v7.2.3, Copyright (c) 1999-2018, by Zend Technologies

Here's the sample script I created to replicate the behavior:

<?php
$example = [
    [
        'a' => 'A',
        'b' => 'B',
    ],
    [
        'c' => 'C',
        'd' => 'D',
    ],
    [
        'e' => 'E',
        'f' => 'F'
    ]
];

foreach ($example as &$letters) {
    foreach ($letters as &$letter) {
        // Do nothing
    }
}

print_r($example);

foreach ($example as $letters) {
    print_r($letters);
}
?>

And here's the output:

Array
(
    [0] => Array
        (
            [a] => A
            [b] => B
        )

    [1] => Array
        (
            [c] => C
            [d] => D
        )

    [2] => Array
        (
            [e] => E
            [f] => F
        )

)
Array
(
    [a] => A
    [b] => B
)
Array
(
    [c] => C
    [d] => D
)
Array
(
    [c] => C
    [d] => D
)

The unexpected behavior is that the second-to-last element replaces the last element only during the second iteration over the array. This is not seen, though, when printing the whole array.

This only happens if there's a nested foreach loop where the values are used by reference. The length of the array does not seem to affect the behavior; the last element is always replaced with the second-to-last element.

Additionally, it appears that the problem is resolved if an unset($letters); statement is added before the last foreach loop.

Is this intentional, or a bug with PHP?

EDIT: See duplicate question above, and specifically this article, which demonstrates visually why the behavior is the way that it is.

Answer 1

Because you're leaving references just lying around waiting to break things. You need to unset() them after the loop to avoid this.

foreach ($example as &$letters) {
    foreach ($letters as &$letter) {
        // Do nothing
    }
    unset($letter);
}
unset($letters);

Also, you should use var_dump() instead of print_r() as it shows you far more useful information that gives you clues to this and many other problems.

eg: Your above, broken output would look like:

array(3) {
  [0]=>
  array(2) {
    ["a"]=>
    string(1) "A"
    ["b"]=>
    string(1) "B"
  }
  [1]=>
  array(2) {
    ["c"]=>
    string(1) "C"
    ["d"]=>
    string(1) "D"
  }
  [2]=>
  &array(2) {
    ["e"]=>
    string(1) "E"
    ["f"]=>
    &string(1) "F"
  }
}
array(2) {
  ["a"]=>
  string(1) "A"
  ["b"]=>
  string(1) "B"
}
array(2) {
  ["c"]=>
  string(1) "C"
  ["d"]=>
  string(1) "D"
}
array(2) {
  ["c"]=>
  string(1) "C"
  ["d"]=>
  string(1) "D"
}

Which shows that the last element in the array and sub-array are still references which causes problems in the subsequent loop.

Answer 2

I believe foreach ($example as &$letters) sets up a variable $letters that points to the original array for the purpose of altering it (rather than making a copy for iteration which is default behavior) and your next statement foreach ($letters as &$letter) modifies the array because it modifies $letters.

Remove the ampersands as I have here:

$example = [
    [
    'a' => 'A',
    'b' => 'B',
    ],
    [
    'c' => 'C',
    'd' => 'D',
    ],
    [
    'e' => 'E',
    'f' => 'F'
    ]
];

foreach ($example as $letters) {
    foreach ($letters as $letter) {
    // Do nothing
    }
}

print_r($example);

foreach ($example as $letters) {
    print_r($letters);
}

And this will be your output:

Array
(
    [0] => Array
    (
        [a] => A
        [b] => B
    )

    [1] => Array
    (
        [c] => C
        [d] => D
    )

    [2] => Array
    (
        [e] => E
        [f] => F
    )

)
Array
(
    [a] => A
    [b] => B
)
Array
(
    [c] => C
    [d] => D
)
Array
(
    [e] => E
    [f] => F
)

Note that if you absolutely must use the ampersands because you do actually want to make changes to the original array, just removing one ampersand will produce that same output:

foreach ($example as $letters) { // removed the ampersand here
    foreach ($letters as &$letter) {
        // Do nothing
    }
}