Planner Row-Count Errors

Multi-column filter conditions lead to poor selectivity estimate [debatable]

Correlations between columns values aren't recognized


Can't See Through WITH

Generic plans can suck

Selectivity estimates on arbitrary expressions are poor

Join selectivity doesn't know about cross-table correlations

Uncommitted tuples don't affect statistics

Histogram bounds can slide due to new data [Fixed]

Inheritance parents aren't assumed to be completely empty [Fixed]

no stats for WITH RECURSIVE

GROUP BY output has no statistics$2q83$

redundant equality constraints not identified as such

IN/NOT IN estimation doesn't assume array elements are distinct [fixed in 9.2]