Dmitry,

> Sorry, this is my fault. It's on page 102,

102 or 112?

> but only the ALL predicate is mentioned there. The fact that NOT IN
>is equal to <> ALL is implied, but it's not very obvious for
>everybody.

Yes, it will be nice, if this will be exactly documented

> v2.0 de-optimizes both ALL and NOT IN subqueries in order to
>guarantee a consistent result.

In the future NOT IN will use indexes, or it will never use indexes ?

So query of type :
select * from t where c NOT IN (select c from t2)
will NOT use indexes
(because this is select * from t where c<>all(select c from t2)

and :
select * from t where c IN (select c from t2)
this will use index or not ?
(because this is select * from t where c = any(select c from t2)

-Laco.