Hello,
I ecountered strange behaviour of function age(timestamp, timestamp).
Let's say we have two timestamps and we want to compute their age() from
fixed timestamp in past:
SELECT pg_catalog.age('2016-04-30 00:00:01'::TIMESTAMP , '2015-01-01
12:00:00'::TIMESTAMP),
pg_catalog.age('2016-04-29 23:59:59'::TIMESTAMP , '2015-01-01
12:00:00'::TIMESTAMP);
age | age
--------------------------------+--------------------------------
1 year 3 mons 28 days 12:00:01 | 1 year 3 mons 28 days 11:59:59
Results are correct.
If we add '1 day'::interval to original timestamps, both results should be 1
day longer.
Howewer first result is exactly 24 hours longer than it should be:
SELECT pg_catalog.age(('2016-04-30 00:00:01'::TIMESTAMP + '1
day'::INTERVAL), '2015-01-01 12:00:00'::TIMESTAMP),
pg_catalog.age(('2016-04-29 23:59:59'::TIMESTAMP + '1
day'::INTERVAL), '2015-01-01 12:00:00'::TIMESTAMP);
age | age
--------------------------------+--------------------------------
1 year 3 mons 30 days 12:00:01 | 1 year 3 mons 29 days 11:59:59
Am I missing something or is it a bug in function age(timestamp, timestamp)?
I tested this on postgres versions 9.4.6 and 9.5.3 on CentOS 7, using
timezone = 'Europe/Prague' in postgresql.conf.
Regards
Michal Merta
--
View this message in context:
http://postgresql.nabble.com/Wrong-results-of-function-age-timestamp-timestamp-tp5924559.html
Sent from the PostgreSQL - bugs mailing list archive at Nabble.com.