The UTF-8 regex can occasionally fail on very low memory machines. Reduce the amount of memory it uses.

See #32204.



git-svn-id: https://develop.svn.wordpress.org/trunk@32375 602fd350-edb4-49c9-b593-d223f7449a82

src/wp-includes/wp-db.php

@@ -2654,7 +2654,7 @@ class wpdb {
 					';
 				}
 
-				$regex .= '){1,50}                          # ...one or more times
+				$regex .= '){1,40}                          # ...one or more times
 					)
 					| .                                  # anything else
 					/x';