in_array vs isset - performance

Question

Consider following snippets of code:

Example #1

$array = Array(1,2,3,4,5,6,7);
$array_test = Array(3,5,4,7,3,6,7,8,8,9,3);

foreach($array_test as $value) {
   if(in_array($value, $array)) {
       // do some magic here
   }
}

Example #2

$array = Array(1,2,3,4,5,6,7);
$array_test = Array(3,5,4,7,3,6,7,8,8,9,3);

$array_index = Array();
foreach($array as $value) {
    $array_index[ $value ] = true;
}

foreach($array_test as $value) {
   if(isset($array_index[ $value ])) {
       // do some magic here
   }
}

Obviously both snippets do the same jobs. In some array samples example #1 is faster than example #2.

I am sure we all were in both situations, however my question is:

• Should I always choose #2?
• When should I choose #1? When size of $array * $array_test is < 10? <100? <1000?
• How to determine which method is better in particular situation?
• Maybe there is some other trick than using temp table $array_index . I don't remember similar situation in other programming languages, everything was ready as-you-go

Please mind about associative keys too.

Someone already asked very similar question:

• what is faster: in_array or isset?

Answer 1

In your second example, you have to construct the "flipped" value of $array before you can use isset() . Btw, you can also use array_flip() for that.

If you can use array keys immediately (without conversion), use isset() because it's obviously much faster than in_array() due to way keys are looked up and because it's a language construct.

If you can't use the keys without conversion, you could consider using in_array() for small arrays. For bigger arrays it might be worthwhile to run a benchmark to determine whether an array conversion step would still be worth it.

Lastly, and depending largely on the situation, you could use one of the array_intersect_ functions as well, mainly to avoid having to loop inside PHP code.

Answer 2

isset will always be faster than in_array because keys are unique and values are not. The fact that keys are unique give optimization possibilities that are impossible with values.