Per field instance default value settings

What makes you think this is a CCK issue ?

This is something we can look at further once we're able to take a pass at how defaults are handled.

Jer

First stab at this. Patch is getting larger than the entire module though. ;)

The module update path for this patch may delay the next release.

Subscribing...

+1 especially for http://drupal.org/node/704014

Cheers

I rerolled #4 for the current 6.x-1.x-dev. The rerolled patch applies cleanly and works pretty well, passing a argument per content type and/or per instance. Legacy tokens (e.g., %nid) are working fine but token module tokens are not working.

Updated #7 to fix Token module replacement and remove the 'token_enabled' concept. Legacy token and Token module token arguments should both work now, per content type or instance.

I'd appreciate some feedback on this patch.

+++ viewfield.module
@@ -37,35 +41,18 @@ function viewfield_field_settings($op, $field) {
     case 'validate':
-      if ($field['force_default'] && $field['multiple']) {
+      if ($field['multiple'] && $field['widget']['force_default']) {

This looks odd -- are the widget settings really available in field settings...?

+++ viewfield.module
@@ -191,74 +159,62 @@ function viewfield_elements() {
+    '#default_value' => $element['#default_value']['vname'],
...
+  // If there is only one option, only show arguments.
+  if (count($options) == 1 && !$is_field_settings_form) {
+    list($key, $label) = each($options);
+    $element['vname']['#access'] = FALSE;
+    $element['vname']['#default_value'] = $key;
   }

Why is the #default_value changed to the first value of #options here? That doesn't look correct to me.

+++ viewfield.module
@@ -191,74 +159,62 @@ function viewfield_elements() {
+  if (module_exists('token')) {
+    $element['vargs']['#description'] .=
+      '<br />' . t('Available tokens: %nid for the id of the current node, %author for the node author, %viewer for the viewing user, or any token from the placeholder token list.')
...
+  else {
+    $element['vargs']['#description'] .= '<br />' . t('Available tokens: %nid for the id of the current node, %author for the node author, and %viewer for viewing user.');
+  }

Let's not duplicate the code for this #description

Powered by Dreditor.

-      if ($field['force_default'] && $field['multiple']) {
+      if ($field['multiple'] && $field['widget']['force_default']) {

Yes, this is wrong. Reverted to original (pre-patch) code.

The code in the patch:

  $element['vname'] = array(
    '#type' => 'select',
    '#title' => $element['#title'],
    '#options' => $options,
    '#default_value' => $element['#default_value']['vname'],
    '#required' => $element['#required'],
    '#access' => $is_field_settings_form || !$field['widget']['force_default'],
    '#description' => $element['#description'],
  );
  // If there is only one option, only show arguments.
  if (count($options) == 1 && !$is_field_settings_form) {
    list($key, $label) = each($options);
    $element['vname']['#access'] = FALSE;
    $element['vname']['#default_value'] = $key;
  }

is derived from this section in viewfield-6.x-1.x-dev :

    if (count($options) > 1) {
      $element['vname'] = array(
        '#type' => 'select',
        '#options' => $options,
        '#default_value' => $element['#default_value']['vname'],
        '#title' => $element['#title'],
        '#required' => $element['#required'],
        '#description' => $element['#description'],
      );
    }
    else {
      // There's only the one view, so only show the arguments.
      list($key, $label) = each($options);
      $element['vname'] = array(
        '#type' => 'value',
        '#value' => $key,
      );
    }

The logic in the 1-option case remains essentially the same.

Removed duplicated legacy token #description in viewfield_elements().

Patch is working for me

Patch now includes code to migrate global default value settings to instance level.

Patch rerolled for latest dev. Please review and test. Once this patch is committed I'll move to the D7 port.

Oh, we're going to commit this into a new 2.x branch?

Might make sense, yeah.

+++ b/viewfield.install
@@ -112,3 +112,39 @@ function viewfield_update_6001() {
+  $result = db_query(
+    "SELECT * FROM {". content_instance_tablename() ."} nfi ".
+    "LEFT JOIN {". content_field_tablename() ."} nf " .
+    "ON nf.field_name = nfi.field_name " .
+    "WHERE nf.type = 'viewfield' "
+  );

It would make sense to define the results of those functions as variables, to subsequently be able to use a single string for the query (on one line) having the $variables embedded. They also seem to be used in the other query of this update.

+++ b/viewfield.install
@@ -112,3 +112,39 @@ function viewfield_update_6001() {
+    $query_string = "UPDATE {" . content_instance_tablename() . "} SET widget_settings='%s' WHERE field_name='%s' AND type_name='%s'";

Can we rename $query_string to $sql?

Would be nice to use spaces between = assignments in the SQL.

+++ b/viewfield.install
@@ -112,3 +112,39 @@ function viewfield_update_6001() {
+    $update = db_query($query_string, $settings_update['widget_settings'], $settings_update['field_name'], $settings_update['type_name']);
+    $ret[] = array('success' => $update !== FALSE, 'query' => sprintf($query_string, $settings_update['widget_settings'], $settings_update['field_name'], $settings_update['type_name']));

Why don't we use update_sql() here?

Otherwise, the code looks good and ready to be committed. I guess we can deal with any problems in follow-up patches as they arise.

Powered by Dreditor.

Rerolled with recommendations.

http://drupalcode.org/project/viewfield.git/commit/f5dec34