On Wed, Mar 18, 2009 at 2:48 PM, Matthew Weier O'Phinney
<[email protected]> wrote:
> -- till <[email protected]> wrote
> (on Wednesday, 18 March 2009, 12:35 PM +0100):
>> On Tue, Mar 17, 2009 at 7:25 PM, Matthew Weier O'Phinney
>> <[email protected]> wrote:
>> > -- till <[email protected]> wrote
>> > (on Tuesday, 17 March 2009, 05:21 PM +0100):
>
> <snip>
>
>> The other thing I noticed was that it did in my ErrorController too (I
>> tested it -- e.g. url does not exist, assert the status code is 404),
>> but I am using setRawHeader() there.
>>
>> I'll give you a hopefully more detailed example:
>>
>> /**
>> * Renders the response
>> *
>> * @param Zend_Controller_Response_Abstract $response - The response
>> object
>> * @return void
>> */
>> public function render(Zend_Controller_Response_Abstract $response)
>> {
>> $response->setHeader('Content-Type', 'text/html; charset=utf-8',
>> true);
>> $response->setHeader('P3P', 'policyref="/w3c/p3p.xml",CP="NOI
>> DSP COR NID CURi DEVa OUR NOR"', true);
>>
>> $response->sendHeaders();
>> $response->outputBody();
>
> These are the problem right here. Zend_Test makes the assumption that
> the front controller's dispatch() method is the only place that sends
> headers and content back to the client; it can then take advantage of
> the returnResponse front controller setting to prevent that from
> happening, which then lets you make assertions against what the response
> object contains.
>
> So, the moral of the story: never send headers and output manually if
> you want to test your controllers.
I still don't understand -- sorry, maybe I am super-slow. How or
where do I set the header correctly? I need the p3p privacy policy,
etc..
Also, my ErrorController forces a 404 response or a 500 when there's
an issue. I gather this is "not allowed" either, or what do you
propose?
> <snip>
>
>> > > 2) My www/index.php kinda looks like this:
>> > > $app = new Bootstrap('environenment');
>> > > $app->start();
>> > >
>> > > Now, I thought I could duplicate this in my tests like this:
>> > >
>> > > ...
>> > > public function setUp()
>> > > {
>> > > $bootstrap = new Bootstrap('testing');
>> > > $this->bootstrap = array($bootstrap, 'start');
>> > > }
>> > >
>> > > ... but to no avail. Whenever I do that, it returns the "entire" html
>> > > of my frontpage, before anything happens. I'm not ruling out that it's
>> > > my fault of course, but I also don't know how to debug this further.
>> >
>> > After you set that, call:
>> >
>> > parent::setUp();
>> >
>> > Otherwise, it won't bootstrap properly. :-) (This *is* documented and
>> > shown in the examples.)
>>
>> Yeah, sorry -- I got that but I forgot to add it here. ;-( Btw,
>> whenever I use $this->bootstrap($bootstrap, 'start') I get the entire
>> homepage's HTML, when I use $this->bootstrap = '/path/www/index.php';
>> I get further.
>
> Odd -- I've used both, and both have worked without issue for me. I may
> need to see the code to better understand what might be at play.
>
> <snip>
>
>> I've discovered the Zend_Session::$_unitTestEnabled flag, but why I am
>> asking is, that, e.g. my tearDown() looks like this:
>>
>> public function tearDown()
>> {
>> Zend_Session::namespaceUnset('fooSession');
>> Zend_Session::expireSessionCookie();
>> Zend_Session::stop();
>> Zend_Session::destroy(true);
>>
>> unset($this->bootstrap);
>> }
>>
>> Yet, all consecutive tests fail because it says:
>> Zend_Session_Exception: A session namespace object already exists for
>> this namespace ('fooSession'), and no additional accessors (session
>> namespace objects) for this namespace are permitted.
>>
>> I'm opening fooSession in my bootstrap, but from what I gather my
>> tearDown() doesn't get rid off everything.
>>
>> That's another thing I've been wrestling with. ;-)
>
> Get rid of your tearDown() method. The bootstrap() method calls reset(),
> which resets the $_SESSION superglobal as well as Zend_Session. What
> you've done in your tearDown() is redundant.
>
> What happens if you remove it?
Well, nothing -- the error stays. I've tried all kind of things in
"tearDown()" because I got the error about the namespace. I wouldn't
mind getting rid off it if it worked without it.
> (...)
Till
