"Paul" <magamos@mail.ru> writes:
> But there is not such thing in PostgreSQL as empty set as "IN ()" that must be
> false, because nothing element may be found in empty set.
> And I thought that instead of "IN ()" I could use "IN (NULL)", but I
> was failed and result was NULL and not FALSE. :(
NULL is not an alternative spelling for an empty set.
You could get an empty IN set by using a sub-select yielding no rows,
for example
regression=# select 1 in (select 1 where false);
?column?
----------
f
(1 row)
regression=# select 1 not in (select 1 where false);
?column?
----------
t
(1 row)
regards, tom lane