From: Michael Paesold <mpaesold@gmx.at>
To: Boris Klug <boris.klug@control.de>
To: Tomasz Myrta <jasiek@klaster.net>
To: Hannu Krosing <hannu@tm.ee>
Cc: pgsql-performance@postgresql.org
Subject: Re: Unions and where optimisation
Date: Wed, 8 Jan 2003 16:48:27 +0100
Message-ID: <013001c2b72d$658ee4b0$3201a8c0@beeblebrox> (raw)
References: <200301081425.48597.boris.klug@control.de>
<1042041735.3237.1.camel@huli>
<3E1C366B.5080406@klaster.net>
<200301081536.33211.boris.klug@control.de>
Boris Klug <boris.klug@control.de> wrote:
> > Hannu, does it work?
> > Few months ago I lost some time trying to create this kind of query and
> > I always got error, that subselect doesn't knows anything about upper
> > (outer?) table.
>
> It does not work on my PostgreSQL 7.2.x
>
> Get the same error like you: "relation rk does not exist"
>
> Also the disadvantage of this solution is that the speed up is bound to
> queries for the ordernr. If a statement has a where clause e.g. for a
> timestamp, the view is still slow.
>
> Does PostgreSQL not know how to move where clause inside each select in a
> union?
Hi Boris,
As far as I know, this has first been "fixed" in 7.3. I think it was Tom who
improved the optimizer to push the where clause into the selects of a union
view. I've done a test...
create view test as
select updated, invoice_id from invoice
union all
select updated, invoice_id from inv2
union all
select updated, invoice_id from inv3;
... and it seems to work (postgresql 7.3 here):
billing=# explain select * from test where invoice_id = 111000;
QUERY PLAN
----------------------------------------------------------------------------
----------------
Subquery Scan test (cost=0.00..413.24 rows=114 width=12)
-> Append (cost=0.00..413.24 rows=114 width=12)
-> Subquery Scan "*SELECT* 1" (cost=0.00..6.00 rows=1 width=12)
-> Index Scan using pk_invoice on invoice (cost=0.00..6.00
rows=1 width=12)
Index Cond: (invoice_id = 111000)
-> Subquery Scan "*SELECT* 2" (cost=0.00..203.62 rows=57
width=12)
-> Index Scan using idx_inv2 on inv2 (cost=0.00..203.62
rows=57 width=12)
Index Cond: (invoice_id = 111000)
-> Subquery Scan "*SELECT* 3" (cost=0.00..203.62 rows=57
width=12)
-> Index Scan using idx_inv3 on inv3 (cost=0.00..203.62
rows=57 width=12)
Index Cond: (invoice_id = 111000)
(11 rows)
I hope this is helps. Can you upgrade to 7.3.1? I really think the upgrade
is worth the effort.
Best Regards,
Michael Paesold
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Reply to all the recipients from your mail client; the link fills in
the headers below: reply via email
Some clients, webmail especially, drop In-Reply-To and References from
a link, and the reply then starts a new thread: use the mbox below.
To: pgsql-performance@postgresql.org
Cc: mpaesold@gmx.at, boris.klug@control.de, jasiek@klaster.net, hannu@tm.ee
Subject: Re: Unions and where optimisation
In-Reply-To: <013001c2b72d$658ee4b0$3201a8c0@beeblebrox>
References: <200301081425.48597.boris.klug@control.de> <1042041735.3237.1.camel@huli> <3E1C366B.5080406@klaster.net> <200301081536.33211.boris.klug@control.de> <013001c2b72d$658ee4b0$3201a8c0@beeblebrox>
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
This inbox is served by DDX for PostgreSQL; see mirroring instructions
for how to clone and mirror all data and code used for this inbox