Description:
------------
parse_url does not parse 'scheme:' correctly.

--------
From http://www.ietf.org/rfc/rfc1738.txt:

5. BNF for specific URL schemes

[snip]

genericurl     = scheme ":" schemepart

[snip]

schemepart     = *xchar | ip-schemepart

--------

In other words, 'scheme:' is a complete and valid URL; the empty string is a valid schemepart.

parse_url ('scheme:') claims that the *host* is 'scheme'.  This is incorrect.  The *scheme* is 'scheme'.

Thanks!


Reproduce code:
---------------
var_dump (parse_url ('scheme:'));


Expected result:
----------------
array(1) {
  ["scheme"]=>
  string(6) "scheme"
}


Actual result:
--------------
array(1) {
  ["host"]=>
  string(6) "scheme"
}