Re: Statistical aggregates with intervals

Hi,

On 24 August 2012 07:39, Christopher Swingley <cswingle(at)gmail(dot)com> wrote:
> I don't know why, but you could convert 'interval' into something else
> where all the functions work:
>
> CREATE OR REPLACE FUNCTION interval_to_seconds(interval)
> RETURNS double precision AS $$
> SELECT (extract(days from $1) * 86400)
> + (extract(hours from $1) * 3600)
> + (extract(minutes from $1) * 60)
> + extract(seconds from $1);
> $$ LANGUAGE SQL;

Looks complicated. You can extract 'epoch':
db=# select now() - (now() - interval '1 day');
?column?
----------
1 day
(1 row)

db=# select extract(epoch from (now() - (now() - interval '1 day')));
date_part
-----------
86400
(1 row)

--
Ondrej Ivanic
(ondrej(dot)ivanic(at)gmail(dot)com)