> Greg Stark wrote:
>> This gets the right semantics but without the debugging info of a list of
>> lockers. Other than debugging the only advantage I see to having the list of
>> lockers is for deadlock detection. Is that absolutely mandatory?
No, deadlock detection is not optional.
Mike Mascari <mascarm@mascari.com> writes:
> What happens if a backend is killed and never decrements its reference
> count?
Even if it's not killed, how does it know to decrement the reference
count? You still need a list of all locked tuples *somewhere*. Perhaps
a technique like this would allow the list to not be in shared memory,
which is helpful, but it's far from an ideal solution.
regards, tom lane