Comment 3 for bug 479743

Revision history for this message
fmpwizard (diego-fmpwizard) wrote : Re: [Bug 479743] Re: update with sub select produces invalid results from transaction_reader/transaction_log

Hi,

On Tue, Nov 10, 2009 at 10:19 AM, Joe Daly <email address hidden> wrote:
> ah yes, I knew that. Ill update the test case with a primary key and
> push it up. It does work with the primary key. Theres really not a great
> place to put a error message if you allow tables without primary keys.
> If someone has this error they would have to alter the table and restart
> replication basically to get the master and slave in sync again?  A long
> term fix may be to verify the schema is suitable for replication.
>

I think there is another thread on the mailing list addressing this
but, could Drizzle handle the missing primary key in a similar way
innodb does?
What I mean is, if you have an innodb table without a primary key,
innodb creates one for you internally (which I believe the user has no
access to).

IMHO, as an end user, this is cleaner than having to add a primary key
to all my tables just because Drizzle replication needs it :)

Thanks

          -Diego

--
Diego Medina
Web Developer
http://www.fmpwizard.com