Roman,

>> Can you explain why is a necessity to accept that people do
>> things against documentation and write stored procedure with RETURNS
>> clauses but without SUSPEND statements?

R> because PSQL compiler does not complain when it sees following code:

R> CREATE PROCEDURE test(a INTEGER) RETURNS (b INTEGER)
R> AS BEGIN
R> b = a;
R> END

R> Since compiler does not complain, this procedure is fully legitimate.
R> But when I check whether it has output params and use "SELECT * FROM
R> test(1)" I get empty result set, not a result set with 1 row and 1 column.

OK. I see the point. I fear you're trying to frame into JDBC's {call}
artifact an implementation that doesn't like it very much :-(
I have already said that I personally would welcome with favour a
change that let execute procedure return a result set when
appropriate; others have pointed out that it wouldn't be backward
compatible, yet a third verb seems like poor design to many. I guess
the discussion is stuck in search of that bright idea that'll solve
everything.

Ciao
--
Nando Dessena
http://www.flamerobin.org