On Wed, 2009-11-25 at 09:23 +0100, Pavel Stehule wrote:
> > If SRFs use a tuplestore in that situation, it sounds like that should
> > be fixed. Why do we need to provide alternate syntax involving COPY?
>
> It isn't problem of SRF function design. It allow both mode - row and
> tuplestor.
select * from generate_series(1,1000000000) limit 1;
That statement takes a long time, which indicates to me that it's
materializing the result of the SRF. And there's no insert there.
> This is problem of INSERT statement, resp. INSERT INTO
> SELECT implementation.
If "tmp" is a new table, and "zero" is a table with a million zeros in
it, then:
insert into tmp select 1/i from zero;
fails instantly. That tells me that it's not materializing the result of
the select; rather, it's feeding the rows in one at a time.
Can show me in more detail what you mean? I'm having difficulty
understanding your short replies.
Regards,
Jeff Davis
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
