Dave Smith wrote:
Because performance would be poor on large tables. In order to find
when a table was last modified, MySQL would have to scan every row in
every table. I would also have to modify all my existing tables to
include a timestamp column. The trigger-based solution has a constant
time complexity (in terns of each table's row count) for timestamp
lookups, while your solution appears to be linear (read: long-term
problems down the road). The trigger-based solution also requires no
change to the existing tables' schema. Seems superior to me. You?
--Dave
I agree. That does sound better. I may give that a try on some of my
own databases and tables. I wonder if this falls under one of the
normalization rules I haven't committed to memory beyond level 3...
Brandon Stout
http://mscis.org
/*
PLUG: http://plug.org, #utah on irc.freenode.net
Unsubscribe: http://plug.org/mailman/options/plug
Don't fear the penguin.
*/
