"Albert REINER" <areiner@tph.tuwien.ac.at> writes:
> Is it really reasonable to enforce that the number of years is four
> digits at least?
I believe so. Without that cue it's pretty difficult for the timestamp
parser even to figure out which field is intended to be the year, let
alone whether you'd like 1900 or 2000 added to a two-digit year value.
For example, with the default datestyle:
regression=# select '12-5-28'::timestamp; ?column?
------------------------2028-12-05 00:00:00-05
(1 row)
regression=# select '0012-5-28'::timestamp; ?column?
---------------------0012-05-28 00:00:00
(1 row)
regards, tom lane