Cumulative settings improvements and fixes.

Please check attached patch. It contains the following:

@@ -465,7 +465,7 @@ function nodeaccess_nodereference_node_access_records($node)
- returns always an array, which gives easier logic, and is in accordance with hook documentation;
- has a small improvement, due to NANR_field_settings() now returning only array's.

@@ -564,15 +561,20 @@ function nodeaccess_nodereference_field_settings()
- logic is now clearer IMO;
- returns always an array, which is easier for parsing the resulting data, as in .._access-records() ;
- has some hints for #1630658: Less memory-usage in nodeaccess_nodereference_field_settings()

@@ -603,7 +608,7 @@ function nodeaccess_nodereference_check_cross_access()
- has a small modification to remove the message. Please check if the logic is not broken.

a better patch.

@#3, indeed, we do not need if (!empty($field_data)) { if the function ALWAYS returns an array, perhaps an empty one. I did not have the overview to remove it on other places.

@#4, I thought that this get/set function was a nice place to centralize the data. If done outside, you must make sure it is called on all set-statements.
@#5, I tried something like

$default_data = ... ;
$user_data = ... ;
$user_data = $default_data + $user_data ; // perhaps the $user_data contains extra unforeseen data, 
if ($user_data == $default_data) {...}

But I couldn't easily get a nice set of default_data in the fist place..

In some other module, we had $default_settings (but it was a simple array - this is more complex). You can mimic the submit data, and you can use it to set the field_settings, too

Nice, such a feuilleton.. :-)
I won't be able to test until next tuesday, so I'll await more episodes...

I think this works fine. Thanks.

Addition 1: I couldn't resist creating an update function. It removes superfluous data in your variable.
The data will be deleted when you save a Node type, but who knows when that happens..
See attached patch (sorry, no git)

2 more suggestions for better readibility of the code:
- move the function nodeaccess_nodereference_field_delete_instance($instance) underneath the ..._update_instance()
- rename internal functions to _nodeaccess_nodereference_...() . This way it is easier to make the difference between hooks and own/internal functions.

Suppose a node type with node reference, but without NANR settings.
The code below invalidates the cache, which should not be necessary for a 'normal' node type.

function nodeaccess_nodereference_node_access_records($node) {
  $grants = array();
  $field_data = nodeaccess_nodereference_field_settings($node->type);
  if (!empty($field_data)) {
    ...  // <= Processes a node with relevant NANR settings
  }

  // Flag that the cache is not to be relied upon.
  variable_set('nodeaccess_nodereference_cache_needs_clear', TRUE); // <= Why is the cache invalid?

  return $grants;
}

We need these lines :

+     if (isset($prior_instance['settings']['nodeaccess_nodereference']) ) {
        $old_settings = nodeaccess_nodereference_reduce_variable($prior_instance['settings']['nodeaccess_nodereference']);
+     }
+     else {
+       $old_settings = FALSE;
+     }
      $new_settings = nodeaccess_nodereference_reduce_variable($instance['settings']['nodeaccess_nodereference']);

I still get the following, using code of this morning:
Notice: Undefined index: unused in nodeaccess_nodereference_node_access_records() (line 507

and this:
Warning: Invalid argument supplied for foreach() in nodeaccess_nodereference_reduce_variable() (line 624 of

recap:
#19: the attached patch for an update function is necessary, to clean up existing saved data and avoid messages.
#20: variable_set('nodeaccess_nodereference_cache_needs_clear', TRUE); is OK, according to #22-24, but I'm not convinced, yet.
#21: is solved with #25
#26: Notice: Undefined index: unused does not occur after activating the updated function, or resaving the field-data
#27-#28: Warning: Invalid argument supplied for foreach() idem
#29: if ($data['unused'] && isset($data['referenced'])) idem

I now see that my update function 7104 is too complicated, and may adopt the style of 7103.

it's Latin, 'the same as', 'dito'

Seems OK, do you need me to test it?

1 typo: "Reduced settings to essentials."

Daniel, will you copy the changes to node access user reference?

Both in NANR and NAUR we need this :

-     $new_settings = nodeaccess_userreference_reduce_variable($instance['settings']['nodeaccess_XXXXreference']);
+     if (isset($instance['settings']['nodeaccess_XXXXreference'])) {
+       $new_settings = nodeaccess_userreference_reduce_variable($instance['settings']['nodeaccess_XXXXreference']);
+     }
+     else {
+       $new_settings = array();
+     }

An error occurs when you add a new XXXreference field and save without settings, or when you remove all settings from an existing field.

Daniel, thanx for the coöperation.

PS. You might want to synchronise the .info files, too.

Node access node reference 7.x-1.21+5-dev Provides per node access control with Node reference fields.
Node access user reference 7.x-3.8+3-dev Provides node access through user reference fields.