Dennis Bjorklund <db@zigo.dhs.org> writes:
> On Wed, 26 May 2004, Tom Lane wrote:
>> I'm not real sure where to document this
>> trick but it seems like we ought to mention it someplace.
> Isn't it better to detect a UPDATE without a where and do that update in
> the same way as the alter table above? Then we don't need to document and
> learn a new non standard way of doing an update.
No, because the locking implications are completely different. I don't
want UPDATE to suddenly decide it needs an exclusive lock on the table
based on the shape of the WHERE clause.
regards, tom lane