Re: Improving count(*) - Mailing list pgsql-hackers

From Tom Lane
Subject Re: Improving count(*)
Date
Msg-id 7711.1132338904@sss.pgh.pa.us
Whole thread Raw
In response to Re: Improving count(*)  (Richard Huxton <dev@archonet.com>)
Responses Re: Improving count(*)  (Alvaro Herrera <alvherre@commandprompt.com>)
List pgsql-hackers
Richard Huxton <dev@archonet.com> writes:
> Might it be possible to apply rule-style rewriting to a clause of an 
> ordinary select query? That is, is it prohibitively expensive to get PG 
> to recognise
>    SELECT count(*) FROM big_table
> and replace it with
>    SELECT sum(summary_count) FROM my_materialised_view

> This should allow you to have where-clauses and apply to a range of 
> cases. What I fear is that checking to see if the rule applies will cost 
>   too much on all those queries where it doesn't apply.

There is already code in the optimizer that does similar rewriting
for min/max queries.   However, that's a hard-wired transformation.
I don't see any very simple way to provide a user-configurable
equivalent.
        regards, tom lane


pgsql-hackers by date:

Previous
From: Tom Lane
Date:
Subject: Re: MERGE vs REPLACE
Next
From: Grzegorz Jaskiewicz
Date:
Subject: order by, for custom types