Thomas Lockhart <lockhart(at)alumni(dot)caltech(dot)edu> writes:
>> Query to reproduce the bug (7.0.3):
>> edge=# select date_part('dow','4/1/2001'::date)::int4;
>> ----------
>> 6
> I'm not seeing this on my 7.0.2 RPM installation, or on my from-cvs
> current sources (dow for April 1 comes up as zero, as you would expect).
I get the right answer from current sources, and the wrong one on 7.0.2.
It's ye olde date-to-timestamp-off-an-hour-at-DST-boundary problem.
Are you not running in a USA timezone?
regards, tom lane