El Jue 28 Ago 2003 11:25, Ignatius Reilly escribió: > Well, it IS logical provided one is aware of the rules of the game, > meaning, as you rightly point out, that MySQL does not reject SQL > statements for incorrect values, but attempts at converting them "at any > price". I have no clue whether it is or not a shortcoming of the > implementation (I do not know what SQL-92 says about this). Therefore data > validation should be done outside MySQL. The thing that IS ilogical is the way MySQL plays the game. What would you think if all of a sudden the "date" command in linux did something like that (try to add some strange date when the given date isn't a valid date)? Now lets suppose this is a good idea, and one would have to program to get the data integrity (data integrity isn't only when the database server fails). Then why do I have to put different data types? Let's use text everywhere! Or if I do use different data types, what if I have a bug? I could have people loading erronous data in the database and not getting an error, and after loading lots of information I find the bug, but see that all the data is bogus (like the numeric example I gave earlier). To state it more clearly, I havn't seen Oracle, Informix or PostgreSQL do things like this, which reminds me of what someone told me once: MySQL is just a FS with an lousy SQL language (and pretty incomplete). Maybe the problem is that most people never read Codd's 12 rules for a RDBMS: http://www.byte.com/art/9406/sec8/art11.htm -- 16:23:01 up 6 days, 8:13, 4 users, load average: 0.16, 0.31, 0.18 ----------------------------------------------------------------- Martín Marqués | mmarques@unl.edu.ar Programador, Administrador, DBA | Centro de Telematica Universidad Nacional del Litoral ----------------------------------------------------------------- -- PHP Database Mailing List (http://www.php.net/) To unsubscribe, visit: http://www.php.net/unsub.php
