Re: Bug in storing Timestamp

On Thu, 21 Apr 2005, Boris Kirzner wrote:

>
> Kris Jurka wrote:
>
> >I am not able to reproduce this problem. The attached test case shows a
> >problem with storing java.sql.Time into a timetz type because the correct
> >offset is not saved, but that's not what you've described. Perhaps it is
> >something specific to your timezone?
> >
> After more investigation I found that the bug I'm talking about can be
> reproduced on timestamp column also.
>
> The configuration is as follows :
> PostgeSQL runs on WindowsXP that has "Automatically adjust clock for dst
> savings" OFF, timezone is (GMT+02:00) Jerusalem
> Client runs on WindowsXP that has "Automatically adjust clock for dst
> savings" OFF , timezone is (GMT+02:00) Jerusalem
>
> The test stores 2004-08-09 20:30:15.50 timestamp into a table and
> immediately fetches it.
> Actual result : the value fetched is 2004-08-09 21:30:15.5
> Expected result : the value fetched is 2004-08-09 20:30:15.5

This is a known bug in the 8.0 driver and seems different than your
original complaint. Being off by an hour (the difference between the two
timezones) is the expected (and incorrect) behavior that I expected to
see. Your original complaint was off by 20:40 which is an unusual number,
are you still seeing that?

Kris Jurka