From: Tom Lane <tgl@sss.pgh.pa.us>
To: James Cloos <cloos@jhcloos.com>
cc: pgsql-sql@postgresql.org
Subject: Re: using possibly null timestamptz columns
In-reply-to: <m3a8eqx4rx.fsf@carbon.jhcloos.org>
References: <m3a8eqx4rx.fsf@carbon.jhcloos.org>
Comments: In-reply-to James Cloos <cloos@jhcloos.com>
	message dated "Thu, 29 Sep 2016 08:35:30 -0400"
Date: Thu, 29 Sep 2016 09:05:48 -0400
Message-ID: <5086.1475154348@sss.pgh.pa.us>
Precedence: bulk
Sender: pgsql-sql-owner@postgresql.org

James Cloos <cloos@jhcloos.com> writes:
> Given a table with a pair of timestamptz columns (lets call them s and e)
> which are typically null, is there a better way to write this where clause
> snippet:

>     where ( s is null or s <= now() ) and ( e is null or e >= now() )

You could try constructing a GIST or SPGIST index on the ranges
tstzrange(s, e), where you'd have to do something to convert null
endpoints to infinities, and then probing with WHERE rangeexpr @> now().

I'm not really sure how well this would perform, but certainly you're
dead in the water as far as doing anything useful with regular btree
indexes.

			regards, tom lane


-- 
Sent via pgsql-sql mailing list (pgsql-sql@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-sql