Обсуждение: BUG #11031: Unexpected interval representations
The following bug has been logged on the website:
Bug reference: 11031
Logged by: Marius Bauer
Email address: marius.bauer@here.com
PostgreSQL version: 9.3.3
Operating system: Amazon RDS
Description:
Please see following query and result:
unitystg=> select a + b as sum, b - a as dif, 2 * a as pro, b / 2 as quo
from (select interval'1 day 15:45:25' as a, interval'2 days 12:34:56' as b)
as foo;
sum | dif | pro | quo
-----------------+-----------------+-----------------+----------------
3 days 28:20:21 | 1 day -03:10:29 | 2 days 31:30:50 | 1 day 06:17:28
(1 row)
Usually I'd expect the day will be adjusted in those cases when the hour is
>= 24 or < 0, but this does not happen.
The interval value itself is correct, just the representation is unexpected.
marius.bauer wrote > The following bug has been logged on the website: > > Bug reference: 11031 > Logged by: Marius Bauer > Email address: > marius.bauer@ > PostgreSQL version: 9.3.3 > Operating system: Amazon RDS > Description: > > Please see following query and result: > > unitystg=> select a + b as sum, b - a as dif, 2 * a as pro, b / 2 as quo > from (select interval'1 day 15:45:25' as a, interval'2 days 12:34:56' as > b) > as foo; > sum | dif | pro | quo > -----------------+-----------------+-----------------+---------------- > 3 days 28:20:21 | 1 day -03:10:29 | 2 days 31:30:50 | 1 day 06:17:28 > (1 row) > > Usually I'd expect the day will be adjusted in those cases when the hour > is >>= 24 or < 0, but this does not happen. > > The interval value itself is correct, just the representation is > unexpected. Unexpected != Bug; this question belongs on -general See http://www.postgresql.org/docs/9.3/interactive/functions-datetime.html Justify_days Justify_hours David J. -- View this message in context: http://postgresql.1045698.n5.nabble.com/BUG-11031-Unexpected-interval-representations-tp5812713p5812727.html Sent from the PostgreSQL - bugs mailing list archive at Nabble.com.
Thanks David, then please take this as an idea to let PostgreSQL call those functions implicitly when an interval is cast to a text, possibly enabling switching this behaviour on and off via a configuration setting. That'd make our lives a little easier. Thanks again! :) -- View this message in context: http://postgresql.1045698.n5.nabble.com/BUG-11031-Unexpected-interval-representations-tp5812713p5812822.html Sent from the PostgreSQL - bugs mailing list archive at Nabble.com.
marius.bauer wrote > Thanks David, then please take this as an idea to let PostgreSQL call > those functions implicitly when an interval is cast to a text, possibly > enabling switching this behaviour on and off via a configuration setting. > That'd make our lives a little easier. Thanks again! :) Not my call but I doubt such a patch would be approved. If you want that behavior use the supplied functions. Enabling a change in this behavior via configuration will never fly and if we make it the default there would be no way to get at the unaligned value. You need a stronger argument than making lives a little easier. Conformance with specs, or at least a majority of other systems, would at least have the inter-operability argument. David J. -- View this message in context: http://postgresql.1045698.n5.nabble.com/BUG-11031-Unexpected-interval-representations-tp5812713p5812836.html Sent from the PostgreSQL - bugs mailing list archive at Nabble.com.