excluded node types should be skipped during counting and indexing from {apachesolr_search_node}

Probably not ideal - we need to add some extra logic to the admin form so we can track when there is a state change I think.

if you change the settings, they formerly excluded nodes will never get in the index.

It might be better to either add another DB column or use the status column.

UPDATE

In that case, you need to delete them if the setting is changed or insert them. Either approach could work.

+  $exclude = array();
+  foreach ($excluded_types AS $type) {
+    if ($type != '0') {
+      $exclude[] = $type;
+    }
+  }

'AS' -> 'as'

use !empty()rather than != '0' which is too implementation specific.

You night also iterate through all existing types and check if they are in the excluded list.

+  $total = db_result(db_query("SELECT COUNT(*) FROM {node} WHERE type NOT IN('".implode(',', $exclude)."') AND status = 1"));

This has code-style problems and shoudl use placeholders ad pass the arguments to the query rather than directly imploding into the query.

Looking at this - it might just be simpler to join to the node table and add an extra WHERE clause, rather than doing all this inserting/deleting.

Looks useful. Read through the code and can test tomorrow.
function _apachesolr_exclude_types needs a code comment

Slight enhancement - add a daily check that we did not fail to delete any excluded nodes. Though I'm a little unsure about the namespaces thing - maybe this check should be in apacehsolr_search for jsut its excluded types?

Looks good.

Thinking more about this - the code to catch any failed delete shoudl not be in the framework. Our one mainn use case for namespaces has been node attachments. So, for example, if I exclude attachments on 'story' nodes that does NOT mean that all 'story' nodes should be deleted from the index.

We tested the patch in #13:
It worked but the apache solr index table never seems to get cleaned out, so it still has info from all the nodes types that have been exlcuded, but were previously indexed. (even after index deletion)

@Jody - that's by design. The latest patch leaves all the nodes in the table, but just excludes them from indexing via the JOIN sql.

better title

actually I think the patch in #13 might be good enough - perhaps we could remove the check empty($old_excluded_types[$type]) here so that admins can re-submit to send delete queries if they fail.

  foreach ($new_excluded_types as $type => $excluded) {
    // Remove newly omitted node types.
    if (!empty($new_excluded_types[$type]) && empty($old_excluded_types[$type])) {
      $solr = apachesolr_get_solr();
      $solr->deleteByQuery("type:$type");
    }
  }

ok, with updated README too.

committed to 6.x