Origionally from UDIG-1233:
Paul Ramsey: "I've experienced the same issue with PostGIS layers, in particular the bc_voting_areas table in the workshop postgis data. RC10 was what I was using."
The postgis case is based on not being able to get an accurate bounds from PostGIS; we need a seperate jira issue here. We use ANALYSIS and random sampling in that order. We would like to reduce this to a database maintenance task ... or ask for a custom function to access the index (which has the information we need).
We will need to take the bounds estimated from PostGIS and try and back out another 20% (at least until an accurate function is available).