"Luke Lonergan" <llonergan@greenplum.com> writes:
> Oops - I missed the point earlier. Start and End are separate attributes so
> this is like an unbounded window in a Start,End space. PostGis provides
> quadtree indexing would provide a terse TID list but you still have the
> problem of how to ensure that the heap tuples being scanned are efficiently
> retrieved, which would only happen if they are grouped similarly to the
> retrieval pattern, right?
Yeah, but I think that's a second-order problem compared to having an
index that's reasonably well matched to the query ...
regards, tom lane