[Tests] Errors with prepareTestedSiblings() lead to non execution of tests

[vjandrea]

There's something wrong with the return value of prepareTestedSiblings() because everytime it's called i get a InvalidArgumentException: Value must be provided. error message.
I have an issue with xdebug and i can't dig deeper into this at the moment, anyway i noticed that if there's an error a test execution is halted, because a $this->assertEquals(true, false) dropped below the assignment is never evaluated.

[franzose]

Strange, I use phpunit, as laravel says, and no errors appear.

[vjandrea]

I have two ClosureTable folders, one inside a laravel project that i update via composer to your repo, one in my documents folder, and in both cases i get the same errors.
htdocs: http://paste.laravel.com/WRv
documents: http://paste.laravel.com/WRw

OSX 10.7.5 / macbookpro
PHP version: 5.4.10 (MAMP)
PHPunit version: 3.7.27

Dependencies installed through composer (identical in both folders)

  illuminate/container 4.0.x-dev ee86c89 
  illuminate/database  4.0.x-dev 64c9e48 
  illuminate/events    4.0.x-dev 2253c8c 
  illuminate/support   4.0.x-dev bec895e 
  nesbot/carbon        1.4.0             A simple API extension for DateTime.

If you need more info let me know!

[vjandrea]

Updated php to 5.5.3, errors persist. I don't know where i should bang my head on. I'm going to setup a vm with vagrant to test on linux.

[vjandrea]

I did further tests on a vm created with vagrant from this repo: https://github.com/bryannielsen/Laravel4-Vagrant and i still get all 34 errors: http://paste.laravel.com/WV4

[franzose]

My setup is Mac OS X 10.7.5 and I don't use virtual machine. PHP is:

PHP 5.4.16 (cli) (built: Jun 15 2013 04:04:59) 
Copyright (c) 1997-2013 The PHP Group
Zend Engine v2.4.0, Copyright (c) 1998-2013 Zend Technologies
    with Xdebug v2.2.2, Copyright (c) 2002-2013, by Derick Rethans

PHPUnit 3.7.21 by Sebastian Bergmann.

[franzose]

Line 544 is where $this->getKey() is called, hm...

[vjandrea]

What makes me think is that before the last call to $this->prepareTestedSiblings(), it stops on line 70, at the end of this:

$child = $page->appendChild(Page::make(array(
            'title' => 'Child Page Test Title',
            'excerpt' => 'Child Page Test Excerpt',
            'content' => 'Child Page Test content',
            'language' => 'en'
        )), 0, true);

or on line 117, at the end of this:

$child1 = $page->appendChild(Page::make(array(
            'title' => 'Test Title',
            'excerpt' => 'Test Excerpt',
            'content' => 'Test content',
            'language' => 'en'
        )), 0, true);

[vjandrea]

Tried on Windows 7 32bit (MINGW32_NT-6.1) as well, PHP 5.5.4 (cli), PHPUnit 3.7.27, i still get all 34 errors:
http://paste.laravel.com/X44

[franzose]

How have you sped, Andea?

[vjandrea]

@franzose, nothing new at the moment, i tried to write other tests for prepareTestedSiblings() but i always get the InvalidArgumentException. I am now studying a bit phpunit API to see if i can make it spit out a more detailed log, because 'Value must be provided' is not really helpful as hint. By the way, did you manage to see the errors in your phpunit?

[franzose]

@vjandrea, and now, you wouldn't believe, I've got the same 34 errors! great. So, now it's a problem of both of us :D

[franzose]

And I know the reason.

[franzose]

@vjandrea, when we call Entity::make(), it tries to guess and set the position if not provided by this: $model->guessPositionOnCreate();. What do we see inside that method? Yes, we see $this->hasSiblings() and $this->lastSibling() calls. Exception is thrown right after $this->hasSiblings() call because... The query is built by buildSiblingsQuery() and buildSiblingsSubquery() methods. The latter requires id of the model. But the model doesn't have it yet because hasn't been saved into the database!

[franzose]

We have to change somehow the way position is guessed.

[vjandrea]

Amazing! Now i know where i should bang my head on! I'm trying to figure out something.

[vjandrea]

So, what if we put a check in hasSiblings() that returns false if the id is not set? I think we should avoid the creation of the query in that case.

[franzose]

@vjandrea, I think that can break things. Now I'm trying to assume that when we make(), we can set position to 0 as well as the depth. Part of errors are gone after that, but still there's 5.

[vjandrea]

@franzose that's great! Do you mind pasting the error dump of the remaining 5, or create a hotfix branch with your edit that i can pull to test?

[franzose]

@vjandrea, aaaaaaaaarrgghhhhhh! Sorry, but it's fucking great. I have solved problems with siblings at last! Sorry, but you'll need to update your repository and merge with your changes.

[vjandrea]

Brilliant job, @franzose!

[franzose]

@vjandrea thanks, however I suppose that there's still problem with the position column. I will discover that a little later.