I don't know much about PostgreSQL, but from a quick google search it seems a "partial" index is a secondary index maintained on a subset of the data (partitioned by some sort of where clause predicate).
Derby currently does not support this kind of index. Currently only indexes that maintain one-to-one relationship between base table rows and index rows are supported. Note that this restriction is not part of the Storage system. The btree implementation does not maintain the relationship, and thus could be used as is to implement such a feature. The changes necessary would be to the optimizer to understand how to cost and use such an index and to the language system to teach it to maintain such an index. A reasonable building block for such a feature would probably be to first implement some sort of "functional" index, which would still provide a one-to-one relationship between base table and index rows but would index on the values returned by a function as applied to one or more columns of a single row. /mikem Garbage Collector wrote: > Does Derby supports partial indexes like that in PostgreSQL or something > like that? > > Thanks in advance. >
