From: valgog <valgog@gmail.com>
Subject: Re: Key/Value reference table generation: INSERT/UPDATE performance
Date: 22 May 2007 03:35:29 -0700
Organization: http://groups.google.com
Lines: 36
Message-ID: <1179830129.622698.240380@x35g2000prf.googlegroups.com>
References: <1179822183.740891.271130@n59g2000hsh.googlegroups.com>
	<op.tsp32yfscigqcu@apollo13>
Mime-Version: 1.0
Content-Type: text/plain; charset="iso-8859-1"
In-Reply-To: <op.tsp32yfscigqcu@apollo13>
User-Agent: G2/1.0
Complaints-To: groups-abuse@google.com
Injection-Info: x35g2000prf.googlegroups.com; posting-host=62.220.8.122;
	posting-account=VhGBRA0AAAAPUct5VN1tGkfI2QuiqXqC
To: pgsql-performance@postgresql.org

On May 22, 12:14 pm, l...@peufeu.com (PFC) wrote:
> On Tue, 22 May 2007 10:23:03 +0200, valgog <val...@gmail.com> wrote:
> > I found several post about INSERT/UPDATE performance in this group,
> > but actually it was not really what I am searching an answer for...
>
> > I have a simple reference table WORD_COUNTS that contains the count of
> > words that appear in a word array storage in another table.
>
>         Mmm.
>
>         If I were you, I would :
>
>         - Create a procedure that flattens all the arrays and returns all the  
> words :
>
> PROCEDURE flatten_arrays RETURNS SETOF TEXT
> FOR word_array IN SELECT word_array FROM your_table LOOP
>         FOR i IN 1...array_upper( word_array ) LOOP
>                 RETURN NEXT tolower( word_array[ i ] )
>
> So, SELECT * FROM flatten_arrays() returns all the words in all the arrays.
> To get the counts quickly I'd do this :
>
> SELECT word, count(*) FROM flatten_arrays() AS word GROUP BY word
>
> You can then populate your counts table very easily and quickly, since  
> it's just a seq scan and hash aggregate. One second for 10.000 rows would  
> be slow.
>
> ---------------------------(end of broadcast)---------------------------
> TIP 3: Have you checked our extensive FAQ?
>
>                http://www.postgresql.org/docs/faq

good idea indeed! will try this approach.