Date: Fri, 16 Jun 2006 11:12:22 -0400
From: Joe <dev@freedomcircle.net>
Subject: Re: Repetitive code
In-reply-to: <bf05e51c0606160744m5f28243cjf0fd6da11f48404d@mail.gmail.com>
To: Aaron Bono <postgresql@aranya.com>
Cc: pgsql-sql@postgresql.org
Message-id: <4492CA56.5020505@freedomcircle.net>
Organization: Freedom Circle, LLC
MIME-version: 1.0
Content-type: text/plain; charset=ISO-8859-1; format=flowed
Content-transfer-encoding: 7bit
References: <44920AA7.7070709@freedomcircle.net>
	<bf05e51c0606151926n18555a65x28a0156f8bd217cc@mail.gmail.com>
	<4492B248.7040406@freedomcircle.net>
	<bf05e51c0606160744m5f28243cjf0fd6da11f48404d@mail.gmail.com>
User-Agent: Thunderbird 1.5.0.4 (Windows/20060516)

Aaron Bono wrote:
> I haven't stared at your query as long as you have so I may have missed 
> something but it looks like in all the selects you are combining the 
> first column in the select is the column you filter on.  So the the 
> outer query doesn't have to know wiether it is a new or changed row:
> 
> SELECT * FROM (
>     SELECT created as my_date, topic_id, 0, 0, 0, 0 FROM topic
>     WHERE page_type IN (1, 2)
>   UNION
> [snip]
>     SELECT e.updated as my_date, e.topic_id, 1, 1, entry_id, rel_type
>     FROM topic_entry e, topic t
>     WHERE e.topic_id = t.topic_id
>       AND date_trunc('day', e.updated) != e.created
>       AND page_type IN (1, 2)
> ) my_union
> where my_union.my_date >= $dt

Thanks Aaron.  That does look like a great solution, overlooked since 
I'm not that familiar with SELECTs in the FROM clause.  It may even make 
it possible to discard the interim table and do the web page/RSS feed 
directly from the view.

> I would almost be tempted to create a view for each small query and name 
> them something meaningful and then another view that does the union.  It 
> would make the queries easier to understand at least (self documented).

That sounds like a good idea too because schema changes would be 
somewhat insulated by the layered views.

Best regards,

Joe