30.11.2010 19:17, Jim Starkey wrote:

> The algorithm is simple, though the details might not be. When a record
> is fetched "for update" in write committed mode:
>
> 1. Look at the head record version. If the transaction that created
> it is committed, do a dummy update and return.
> 2. If the transaction that created the head version is active, wait
> for it to complete. Then go back to step #1

So do you ignore this scenario:

txn 1: update tab set col = 2 where col = 1
txn 2: select * from tab where col = 1 for update
txn 1: commit

and lock the record which supposedly shouldn't have been locked?

Or, once again, I'm missing something?


Dmitry