Re: to_date/to timestamp going to BC

Thanks Karel,Good call, you guessed it. I was just using my Oracle knowledge of
the to_date and applying it to the results I was expecting in pgsql. Guess I
should not make those assumptions....

-----Original Message-----
From: Karel Zak [mailto:zakkr@zf.jcu.cz]
Sent: Thursday, October 04, 2001 2:42 AM
To: Tom Lane
Cc: Servetar, Jason; pgsql-sql@postgresql.org
Subject: Re: [SQL] to_date/to timestamp going to BC


On Wed, Oct 03, 2001 at 05:14:02PM -0400, Tom Lane wrote:
> jason.servetar@ccgenesis.com writes:
> > Can someone tell me if this is a bug with the date functions or am I
using
> > them incorrectly?
> 
> I get the right thing when I use the right format:
> 
> regression=# select dt, to_timestamp(dt, 'FMMonth dd, yyyy') from
test_date;
>        dt       |      to_timestamp
> ----------------+------------------------
>  March 11, 1997 | 1997-03-11 00:00:00-05
> (1 row)
> 
> However, I'd agree that this shows a lack of robustness in to_timestamp;
> it's not objecting to data that doesn't match the format.
The manual is transparent about this. I can add feauture that will
check everythig, but users who knows read manual and use already 
debugged queries will spend CPU on non-wanted code. 

Hmm.. I look at Oracle, and it allows parse queries like:

SVRMGR> select to_date('March 11, 1997', 'Month dd, yyyy') from dual;
TO_DATE('
---------
11-MAR-97
1 row selected.

.. well, I add it to my TODO for 7.3 (I plan rewrite several things 
in to_* functions).
Karel

-- Karel Zak  <zakkr@zf.jcu.cz>http://home.zf.jcu.cz/~zakkr/C, PostgreSQL, PHP, WWW, http://docs.linux.cz,
http://mape.jcu.cz