drupal_static_reset is broken

If this would be wrong, then a lot of tests would fail. So it seems this is almost RTBC.

+++ includes/bootstrap.inc	2009-11-07 09:30:56 +0000
@@ -2066,31 +2066,34 @@ function registry_rebuild() {
   if ($reset) {
...
+    if (array_key_exists($name, $default)) {
+      $data[$name] = $default[$name];
     }
     else {
...
+      // Reset was called before a default is set and yet a variable must be
+      // returned.
+      return $data;
+    }    
   }

Could we add some inline comments here as well?

(Please also note the trailing whites-pace.)

+++ includes/bootstrap.inc	2009-11-07 09:30:56 +0000
@@ -2066,31 +2066,34 @@ function registry_rebuild() {
+  elseif (!array_key_exists($name, $data)) {
+    $default[$name] = $data[$name] = $default_value;
   }

Inline comments here, too, as this seems to be the main static/reference handling.

I'm on crack. Are you, too?

Alright, I'm more happy with this.

Looks good to me too. Committed.

drupal_static_reset() is still broken when called globally (without passing a $name parameter). This was the case prior to this issue, but I'm attaching the patch here, since it's related to this issue. The bug hasn't manifested yet, because we have no places within core where drupal_static_reset() is called without passing $name. This patch includes a test that fails unless this patch is applied.

Yes.

Cross-linking this to #601584: setUp() function for unit and web test cases should reset all resettable statics, because that issue now includes this same fix. If that issue lands first, then this one can be marked "fixed" again as per #14. On the other hand, if this issue lands first, then that one will need to be re-rolled.

Committed to HEAD with some minor comment/whitespace clean-up.

Awesome! So, now that global drupal_static_reset() isn't broken, can anyone here help get #601584: setUp() function for unit and web test cases should reset all resettable statics to RTBC level?