Additionally (and I haven't been in low-level Lucene for a LONG time
so caveat emptor), in general ID types should not be tokenized. I'd
model my field definition after the Lucene StringField class. It looks like
you can avoid defining a FieldType altogether and just use
StringField.TYPE_STORED for
You can check following possibilities :
- doc.id.toString() return same string for all of your documents.
- your tokenstream for doc.id.toString() return same token for all of your
documents. Ex: id "1 23" -> {"1","23"}. id "1 56" -> {"1","56"}. So when
you delete document by using iw.deleteDocumen
