Description:
------------
PCRE has supported Perl's /n modifier since Version 10.30 14-August-2017, and has supported its behaviour via a NO_AUTO_CAPTURE option for many years before that.

It prevents numbered captures appearing in result sets, allowing only named groups through (except 0, which corresponds to Perl's $&). It can make complex patterns more readable, and debugging/output of matches smaller, more semantic, and better suited to iteration.


Test script:
---------------
<?php
preg_match('/(?:a|b)c(?:d|e)/', 'ace', $m1);
preg_match('/(a|b)c(d|e)/n', 'ace', $m2);
preg_match('/(?<prefix>foo|bar)baz(?<suffix>qux)?/', 'foobazqux', $m3);
preg_match('/(?<prefix>foo|bar)baz(?<suffix>qux)?/n', 'foobazqux', $m4);
var_dump($m1, $m2, $m3, $m4);

Expected result:
----------------
array(1) {
  [0]=>
  string(2) "ace"
}
array(1) {
  [0]=>
  string(2) "ace"
}
array(5) {
  [0]=>
  string(9) "foobazqux"
  ["prefix"]=>
  string(3) "foo"
  [1]=>
  string(3) "foo"
  ["suffix"]=>
  string(3) "qux"
  [2]=>
  string(3) "qux"
}
array(5) {
  [0]=>
  string(9) "foobazqux"
  ["prefix"]=>
  string(3) "foo"
  ["suffix"]=>
  string(3) "qux"
}


Actual result:
--------------
Warning: preg_match(): Unknown modifier 'n' in pcre-n.php on line 3

Warning: preg_match(): Unknown modifier 'n' in pcre-n.php on line 5
array(1) {
  [0]=>
  string(2) "ac"
}
NULL
array(5) {
  [0]=>
  string(9) "foobazqux"
  ["prefix"]=>
  string(3) "foo"
  [1]=>
  string(3) "foo"
  ["suffix"]=>
  string(3) "qux"
  [2]=>
  string(3) "qux"
}
NULL