Re: Theoretical Basis for SELECT FOR UPDATE

David Cressey wrote:
[...]
> By SELECT FOR UPDATE, you take out the higher level lock earlier. This
> reduces concurrency, but also reduces deadlocks.
>

SELECT FOR UPDATE is not useful for locking databases and is not implemented there. It's only reason to be, as wrote before, is to emulate repeatable reads in a multiversioning database.

Besides, select for update can create deadlocks as easily as a run-of-the mill update or delete. Received on Fri Sep 30 2005 - 23:44:09 CEST