It strikes me that the btree-disk implementation i.e (package org.apache.xindice.core.filer) used within this product could be made standalone and perhaps moved to the Jakarta-Commons project.
This would effectively provide a very efficient and reliable pure java DBM solution (as opposed to berkeley) similar to the (lot slower) sourceforge jdbm. Please forgive any ignorance or lack of knowledge on my part. Regards Steve Sutton P.S Sorry if this has been posted twice - probs with my ISP...
