--- Original Message ---
From: "Stormblade" <[EMAIL PROTECTED]>
To: Donny Simonton <[EMAIL PROTECTED]>
Cc:
Sent: Tue, 20 Apr 2004 12:59:58 -0400
Subject: RE: Gripe with MySQL


> On Tue, 20 Apr 2004 11:37:12 -0500, Donny Simonton wrote:
> Everybody should remember as well, if you run rm -rf /*.* on your
> server you will delete everything from your server, but linux
> will stay running.  Even though that's not documented either.
>
> If you use a client like PHPMyadmin or one of the other 80
> million that are around you won't have to worry about error
> checking because they have already done it for you.  Now as far
> as your clients/customers, if you don't have error checking in
> yourself, that's your problem not mysql's problem.
>
> Donny
>

One clarification. Database checking does not eliminate the need
for client error checking. The two of these things have nothing to
do with each other. The database should allow for database
protecting checks as it knows nothing and shouldn't know anything
about the client. Whether the client performs their own error
checking or not is irrelevant.

Obviously this is, as I mentioned, a design thing and not everyone
will agree on a design. I prefer the server to allow a database
administration to setup checks to help ensure that data is going in
as was intended. The database administrator in many companies have
nothing to do with writing the clients or the SQL that accesses the
database.

There are some checks that I feel are better located on the server
and there are some that are better located at the client end. When
I tell a field that it should not be blank. I don't mean for the
database to take it upon itself to ensure that it's never ever
blank. If a record comes in to a server, that server should include
a way to tell whether that record is invalid according to the
server's specifications.

This is not a new model and many many applications do this. Both
client and server have responsibilities. We differ on what we feel
those responsibilities are. I feel that I should be able to define
what an invalid record is AND have the server reject that record if
it is invalid.

The server has nothing to do with whether the SQL was valid or
invalid or whether the SQL is sending bad data like the wrong dates
or text fields that are chopped off. That's all client side.
However, the server should know what a valid record looks like and
reject those that are not valid.

Having worked in a distributed environment and using various
middleware I can tell you that in an environment like that it is
crucial that the server be able to tell invalid records from valid
ones and have the option to handle them (reject, accept/fix, log,
etc). Right now MySQL only does an accept/fix type thing. It makes
it a valid record even though it really isn't.
--
Stormblade (Shaolin Code Warrior)
Software Developer (15+ Years Programming exp.)

"The worth of your opinion is in direct proportion to the number of people who have 
asked for it."

My System: http://www.anandtech.com/mysystemrig.html?rigid=1683


--
MySQL General Mailing List
For list archives: http://lists.mysql.com/mysql
To unsubscribe:    http://lists.mysql.com/[EMAIL PROTECTED]

Reply via email to