[Geoserver-devel] [postgis-users] Re: Postgis estimated_extent completely off the mark?

[Mark Cave-Ayland]

On Tue, 2007-03-20 at 08:52 -0700, Paul Ramsey wrote:
> Once again I wonder why we're not pulling the extent off of the top  
> node of the index.

Well, I did post some code to do this once but the problem is that since
visibility information is not stored in the index, you still have to
scan the heap to check whether the index entry is visible for the
current transaction. (Hint: Imagine removing all geometries greater than
a distance from a certain point. All of deleted entries in the "outer"
part of the index will still be in the there, even though most of them
will no longer exist). Also, as strk points out, estimated_extent()
doesn't require an index to work.


Kind regards,

Mark.