Details
-
Bug
-
Status: Open
-
Major
-
Resolution: Unresolved
-
None
-
None
-
None
-
None
Description
Prior to Lucene 5, a single algorithm / implementation could work across both numeric and string indexed fields. With the changes to the FieldCache and the introduction of UninvertingReader, this is no longer true. Code that depends on the FieldCache must now have separate versions for numeric fields vs string/text fields.
This is the root issue behind SOLR-7495, SOLR-7496, and probably others.