Remove DRUPAL_ROOT constant redefinitions from unit tests

This no longer completely applies.

DRUPAL_ROOT should not have been added to the PHPUnit bootstrap.php.

I don't understand why it was added, and even added to a section that clearly states that constants should be removed...?

// Look into removing these later.
define('REQUEST_TIME', (int) $_SERVER['REQUEST_TIME']);
define('DRUPAL_ROOT', realpath(__DIR__ . '/../../'));

The globally defined constants are tampering the environment of all tests, which is bad.

Do we have a @TODO/issue to remove this globalness, for example in the phpunit bootstrap?

+++ b/core/tests/Drupal/Tests/Core/Extension/ThemeHandlerTest.php
@@ -376,9 +376,6 @@ protected function systemListReset() {
 if (!defined('DRUPAL_MINIMUM_PHP')) {
   define('DRUPAL_MINIMUM_PHP', '5.3.10');
 }

I guess this could be moved to \Drupal without any problems? Not sure about early installer though, tbh.

Due to recent work on PHPUnit, I now agree with this change/cleanup. I don't see a way around avoiding DRUPAL_ROOT in the phpunit bootstrap.php either.

In fact, meanwhile I believe that an APP_ROOT constant like this should actually be supplied universally, centrally, and standardized by Composer's autoloader (which deals with many similar band-aid material like PHP INI settings anyway already). The constant is not really Drupal-specific; it simply points to the base/root directory of the application. I'll try to create an issue for that.

That said, note that there are some unit tests in HEAD that try to avoid DRUPAL_ROOT and thus use excessive relative directory name traversals à la dirname(dirname(dirname(__DIR__))) or __DIR__ . '/../../../../' to locate test fixtures. Would be great to adjust those accordingly here?

Instead of using DRUPAL_ROOT we should now use $container->getParameter('app.root'), so we no longer need this workaround.

This is really just a start for now.

@dawehner: Please note the issue title - this issue is about phpunit tests only. There is no container in unit tests.

Can you create a separate issue for #8? (the proposal is questionable, but not to be discussed here)

Okay.

Re-rolled against HEAD. No change, so RTBC.

Correcting slightly misleading issue title.

Committed and pushed to 8.x. Thanks!