Review 1.3
==========

[x] Executing the tests with PHP 5.3.0RC3-dev (cli) (built: May 31 2009 10:43:59)
    ('--with-curl' '--with-curlwrappers') gives me the following error::

        Notice: Array to string conversion in trunk/Authentication/src/url/url.php on line 222

    This is because the $metadata-Array may also contain additional arrays and
    cannot just be imploded. When curl is used by the http-wrapper, the meta-data
    array looks like::

        array(10) {
          ["wrapper_data"]=>
          array(2) {
            ["headers"]=>
            array(0) {
            }
            ["readbuf"]=>
            resource(106) of type (stream)
          }
          ["wrapper_type"]=>
          string(4) "cURL"
          ["stream_type"]=>
          string(4) "cURL"
          ["mode"]=>
          string(1) "r"
          ["unread_bytes"]=>
          int(0)
          ["seekable"]=>
          bool(false)
          ["uri"]=>
          string(16) "http://yahoo.com"
          ["timed_out"]=>
          bool(false)
          ["blocked"]=>
          bool(true)
          ["eof"]=>
          bool(false)
        }

    The same error accurs if I try to authentificate against a OpenID (2) site in
    an application using ezcAuth.

    Seems to be "fixed" by removing the --with-curlwrappers option from ./configure

    AS: I added a check for the 'headers' sub-array, so it should work in
    both cases, with and without --with-curlwrappers. Both cases are
    untestable in the same test suite though.

[ ] I get some test failures, when running the test suite, as recorded in
    "review-1.3-test-failures.txt". Only some of them seem to be temporary
    issues.

[ ] Why is OpenID 2.0 not default? It does only implement an additional
    discovery mechanism and this way some common providers do not work out of
    the box, but $auth->run just returns false.

    Is there any drawback using 2.0 by default, I do not see? I would not
    consider it a BC break...