Tom Lane wrote:
> Josh Berkus <josh@agliodbs.com> writes:
>> A Hibernate developer pointed out the following odd behavior to me in 8.2.1:
>
>> create table test ( test1 text );
>> create table test2 ( test_col text );
>> create rule test_insert as on insert to test do instead insert into test2
>> values ( NEW.test1 ) RETURNING test2.test_col;
>
>> postgres=# insert into test values ( 'joe' );
>> INSERT 0 1
>
>> ... no RETURNING.
>
> It would surely be quite broken for an INSERT that has *not* got a
> returning clause to spit data at you, don't you think?
>
> What the RETURNING clause in the rule does is let you define the data
> that should be returned if the rewritten INSERT had a returning clause
> to start with.
Sorry - haven't got a CSV download here, or I'd check myself. Does this
just allow an INSERT...RETURNING inside the rule, or could it be
something like:
CREATE RULE ... AS ON INSERT ... DO INSTEAD SELECT f(NEW.test1);
-- Richard Huxton Archonet Ltd