> the current Derby semantics (releasing
the update lock once a next() is executed), isn't very helpful.

Even in read-committed isolation level, it still seems like it
might be useful to be able to control the lock-mode on the current
row. If I have an read-committed transaction which reads through
many rows, and chooses to update a subset of them, being able
to communicate to the DB that I'd like an update-mode lock when
I read the record, to be either (a) released when I do Next or
(b) promoted to exclusive-mode if I happen to update this record,
seems like a useful behavior to me.

That is, I think that update-mode locks can be useful in
reducing deadlock, and I think that's true even when running at
read-committed isolation level.

thanks,

bryan

Reply via email to