strncmp

A note not included in the documentation:
int strcmp ( string $str1 , string $str2 )
Returns < 0 if str1 is less than str2; > 0 if str1 is greater than str2, and 0 if they are equal.
My addendum:
If str1 and str2 are not equal, and str1 is a sub-string of str2 or vise versa. The returned int value will be negative or positive indicating how many characters the difference is between the two strings in absolute terms.
Example:
<?php
$str1 = "phpaaa";
$str2 = "php";
echo strcmp($str1, $str2); // 3
?>

since str2 = "php" is a sub-string of str1 = "phpaaa" and "phpaaa" is greater than "php" the returned value is positive and is 3 indicating how many characters the difference is between the two strings.
If you replace the value of str1 with str2 the result will be -3 (negative) but still indicates the absolute difference which is 3

I ran the following experiment to compare arrays.
1 st - using (substr($key,0,5 == "HTTP_") & 2 nd - using (!strncmp($key, 'HTTP_', 5))
I wanted to work out the fastest way to get the first few characters from a array
BENCHMARK ITERATION RESULT IS:
if (substr($key,0,5 == "HTTP_").... - 0,000481s
if (!strncmp($key, 'HTTP_', 5)).... - 0,000405s
strncmp() is 20% faster than substr() :D
<?php
// SAMPLE FUNCTION
function strncmp_match($arr)
{
foreach ($arr as $key => $val)
 {
 //if (substr($key,0,5 == "HTTP_")
 if (!strncmp($key, 'HTTP_', 5)) 
 {
 $out[$key] = $val;
 }
 }
return $out;
}
// EXAMPLE USE
?><pre><?php
print_r(strncmp_match($_SERVER));
?></pre>
will display code like this:
Array
(
 [HTTP_ACCEPT] => XXX
 [HTTP_ACCEPT_LANGUAGE] => pl
 [HTTP_UA_CPU] => x64
 [HTTP_ACCEPT_ENCODING] => gzip, deflate
 [HTTP_USER_AGENT] => Mozilla/4.0 
 (compatible; MSIE 7.0;
 Windows NT 5.1;
 .NET CLR 1.1.4322;
 .NET CLR 2.0.50727)
 [HTTP_HOST] => XXX.XXX.XXX.XXX
 [HTTP_CONNECTION] => Keep-Alive
 [HTTP_COOKIE] => __utma=XX;__utmz=XX.utmccn=(direct)|utmcsr=(direct)|utmcmd=(none)
)

I just want to highlight that (at least on php7), when testing for the existence of a string in the beginning of another string you should consider using substr or strpos (if performances is an issue).
Here is a small benchmark (for what it's worth):
<?php
$n = 'abcd';
$l = strlen($n);
$haystack0 = base64_encode(random_bytes(128));
//heat
$r = 1;
for ($i = 0; $i < 100000000; $i++)
 $r += $r * $r % 10000;
//tests
$k = 30000000;
$res = array();
foreach (array('found' => $n . $haystack0, 'not-found' => strrev($n) . $haystack0) as $f => $haystack) {
 $m = microtime(true);
 for ($i = 0; $i < $k; $i++)
 !strncmp($haystack, $n, $l) && $r++;
 $res["strncmp-$f"] = -$m + ($m = microtime(true));
 for ($i = 0; $i < $k; $i++)
 (strpos($haystack, $n) === 0) && $r++;
 $res["strpos-$f"] = -$m + ($m = microtime(true));
 for ($i = 0; $i < $k; $i++)
 (substr($haystack, 0, $l) === $n) && $r++;
 $res["substr-$f"] = microtime(true) - $m;
}
//print
asort($res);
print_r($res);
echo "\n$r"; // makes sure no auto-optimization occurs
?>

This outputs:
<?php /*
 [strpos-found] => 1.3313138484955
 [substr-not-found] => 1.4832630157471
 [substr-found] => 1.6976611614227
 [strpos-not-found] => 2.0043320655823
 [strncmp-not-found] => 2.0969619750977
 [strncmp-found] => 2.3616981506348
*/ ?>

For checking matches at the beginning of a short string, strpos() is about 15% faster than strncmp().
Here's a benchmark program to prove it:
<?php
$haystack = "abcdefghijklmnopqrstuvwxyz";
$needles = array('abc', 'xyz', '123');
foreach ($needles as $needle) {
 $times['strncmp'][$needle] = -microtime(true);
 for ($i = 0; $i < 1000000; $i++) {
 $result = strncmp($haystack, $needle, 3) === 0;
 }
 $times['strncmp'][$needle] += microtime(true);
}
foreach ($needles as $needle) {
 $times['strpos'][$needle] = -microtime(true);
 for ($i = 0; $i < 1000000; $i++) {
 $result = strpos($haystack, $needle) === 0;
 }
 $times['strpos'][$needle] += microtime(true);
}
var_export($times);
?>

if length is 0 regardless what the two strings are, it will return 0
<?php
strncmp("xybc","a3234",0); // 0
strncmp("blah123","hohoho", 0); //0
?>