Message-ID: <3DB6B41C.8090409@klaster.net>
Date: Wed, 23 Oct 2002 16:37:16 +0200
From: Tomasz Myrta <jasiek@klaster.net>
User-Agent: Mozilla/5.0 (Windows; U; Win 9x 4.90; PL; rv:1.1) Gecko/20020826
MIME-Version: 1.0
To: Carlos Sousa <carlosdsousa@hotmail.com>
Cc: pgsql-sql@postgresql.org
Subject: Re: problem with a query
References: <F94LAPDVXLuSbh29cM300000459@hotmail.com>
Content-Type: text/plain; charset=ISO-8859-2; format=flowed
Content-Transfer-Encoding: 8bit

Użytkownik Carlos Sousa napisał:
> hi to all
> 
> i need to do a query to a small postgres database
> my best efort to the pretended query is in the attached file 'query.txt'
> but i need that some rows don't be selected of thouse that the query 
> returns.
> 
> i send in another attached file 'result.txt' with the query result. 
> looking at content of the file you will see some notes at the end of the 
> table (#..)
> 
> the database as the data to enable the constrution of a school class 
> schedule.
> 
> In the 'result.txt' all the rows with max=0 were the fisrt to be 
> inserted do the database. each row represent as example math pratical 
> starting at 8:30 for 1:30 at room 3.15 in the bilding CP
> 
> the rows with max!=0 (1,2,3,...) mean that content of an insert with 
> max=0 was changed [ex diferent time (fiel to_char) or room (field sala) 
> or bilding (field edificio)of the school class]. row with max=1 is a 
> substitution to a row with max=0, row with max=2 substitutes row with 
> max=1 and ...
> 
> my objective is to obtain all the rows that were not substituted and the 
> ones that represent the last alteration to the shedule
> 
> i hope the problem was understod
I'm not sure, but you want to get data for each room (field "sala") with 
  the biggest field "indice"?
you can do this with query:
select distinct on (indice)
  ...
from
  ...
order by sala,indice desc;

Result of this query is - one row per one "sala". If there are more then 
1 records in database with the same "sala", you get record with biggest 
"indice". You don't need to use max(indice) and group by...

Tomasz Myrta