Re: Much Ado About COUNT(*)

Tom Lane wrote:
> Bruce Momjian <pgman(at)candle(dot)pha(dot)pa(dot)us> writes:
> > My basic idea was to keep a status bit on each index entry telling it if
> > a previous backend looked at the heap and determined it was valid.
>
> Even if you could track the tuple's committed-good status reliably,
> that isn't enough under MVCC. The tuple might be committed good, and
> seen that way by some other backend that set the bit, and yet it's not
> supposed to be visible to your older transaction. Or the reverse at
> tuple deletion.

I mentioned that:

> (Oh, and you could only update the bit when all active transactions
> are newer than the creation transaction so we know they should all see
> it as visible.)

--
Bruce Momjian | http://candle.pha.pa.us
pgman(at)candle(dot)pha(dot)pa(dot)us | (610) 359-1001
+ If your life is a hard drive, | 13 Roberts Road
+ Christ can be your backup. | Newtown Square, Pennsylvania 19073