Re: Mysql date/time, was Re: Press Release -- Just

Oh, that is interesting... But that is not a datetime. Timestamp (in case you don't know, though you probably do already) will automatically update itself when the row is edited, but it only works on the first one. I learned that the hard way. They consider this a "feature" as well, but it looks intentional...

-J

-----Original Message-----
From: Richard Huxton [mailto:dev(at)archonet(dot)com]
Sent: Wednesday, November 20, 2002 5:48 AM
To: Jason Hihn
Cc: pgsql-advocacy(at)postgresql(dot)org
Subject: Mysql date/time, was Re: [pgsql-advocacy] Press Release -- Just
Waiting for Tom

On Tuesday 19 Nov 2002 2:43 pm, Jason Hihn wrote:
> Interesting treatment of Date/Time?

+-------+---------------+------+-----+---------+-------+
| Field | Type | Null | Key | Default | Extra |
+-------+---------------+------+-----+---------+-------+
| a | int(11) | YES | | NULL | |
| b | timestamp(14) | YES | | NULL | |
+-------+---------------+------+-----+---------+-------+
2 rows in set (0.00 sec)

mysql> INSERT INTO foo VALUES (1,now());
Query OK, 1 row affected (0.00 sec)

mysql> INSERT INTO foo VALUES (2,null);
Query OK, 1 row affected (0.01 sec)

mysql> INSERT INTO foo VALUES (3,'abcd');
Query OK, 1 row affected (0.00 sec)

mysql> INSERT INTO foo VALUES (4,'');
Query OK, 1 row affected (0.00 sec)

mysql> SELECT * FROM foo;
+------+----------------+
| a | b |
+------+----------------+
| 1 | 20021120104314 |
| 2 | 20021120104326 |
| 3 | 00000000000000 |
| 4 | 00000000000000 |
+------+----------------+
4 rows in set (0.01 sec)
mysql> select version();
+-----------+
| version() |
+-----------+
| 3.23.49 |
+-----------+

The last case caught me out in real code. The handling of null can't be right,
and case 3,4 store an invalid date/time. Not sure if it's been changed in
more recent versions, but it was considered a feature.

--
Richard Huxton