> On Sunday 15 September 2002 10:11 am, Rod Taylor wrote:
>> On Sun, 2002-09-15 at 03:57, Casey Allen Shobe wrote:
>>> I don't have a clue why it's doing this - has anyone else seen this sort
>>> of behavior, or know why it might be doing it? It shows times wrong, for
>>> instance, it shows 00:04:60 where it should show 00:05:00. See below:
>>
>> There are actually 61 seconds in some minutes. In order to accommodate
>> leap seconds, PostgreSQL allows this to happen -- similarly to how it
>> will also allow 366 days in some years.
True but irrelevant -- PG does not do accounting for leap seconds.
The roundoff bug Casey is looking at is fixed in recent PG versions;
I'd recommend an update to 7.2.2.
regards, tom lane