It more than helps: it opens all kinds of possibilities. So, my question now is: Should I have designed the database with a 'date' and 'time' columns from the beginning (vs. just a timestamp)? Thank you very much, Bill Totman On Sunday 29 July 2007 18:26, you wrote: > I believe casting the timestamp to time (or timetz) and date will do > what you want: > > test=# select current_timestamp, current_timestamp::timetz, > current_timestamp::date; > now | now | now > -------------------------------+--------------------+------------ > 2007-07-29 18:16:49.643542-05 | 18:16:49.643542-05 | 2007-07-29 > (1 row) > > If both date and time are important, I'd recommend keeping them in a > timestamp and decomposing when you need to. Depending on what kinds > of queries are performed, you may also want to look into using > expressional indexes, such as: > > CREATE INDEX timestamptz_col_date_idx ON foo (timestamptz_col::date); > CREATE INDEX timestamptz_col_timetz_idx ON foo > (timestamptz_col::timetz); > > Hope this helps. > > Michael Glaesemann > grzm seespotcode net
Attachment:
pgp2qjro3FIfW.pgp
Description: PGP signature