I'll add the missing part to one of statements in the beginning of my
last mail:
> 3rd is probably quite inaccurate, especially when histograms aren't
> calculated on non-single row predicate columns (again, with bind
> variables is useless). Also, if you want to use it, you have to find
> the right row for the
... rowcount (cardinality). For select queries this is normally the
execution plan row with id = 1 in v$sql_plan, but for update for example, it's
the id=0 row that shows cardinality.
Tanel.
