Aaron Stone wrote:
I'm thinking we should go with LIKE.
This is MySQL 4.0.21 on Gentoo Linux on a Pentium III 866.
mysql> SELECT DISTINCT(physmessage_id) FROM dbmail_messageblks WHERE
messageblk LIKE '%From:%Aaron%';
11089 rows in set (31.17 sec) [cpu hovered around 50%]
mysql> SELECT DISTINCT(physmessage_id) FROM dbmail_messageblks WHERE
messageblk REGEXP '.*From:.*Aaron.*';
11089 rows in set (10 min 44.80 sec) [cpu was pegged solid at 99.9%]
I haven't actually done the benchmark recently, but I think there are
similiar performance differences between like and regex on postgresql.
Use regex only when you cant' get what you need from like.
Matthew
