olgreyfox15 wrote:

> The idea of a time stamp is ok? But has problems based on real world
> scenario.

My suggested implementation is also based on real world scenario!

On the corner case? If the time clocks between multiple

> computers are out of sync? Then you have the risk of the wrong bidder
> being posted as the winner. That is a problem.

As Woody has been already said, since the timestamp values will (should)
all come from the server, there will not be a problem.

> Let me see if I can clarify a couple of things. First? This is for
> benefit auctions, not commercial. (Big difference) The auctioneer is
> no where near the computer system. In fact we do not want them near
> the computer! ;) Data entry is being done by multiple computers and
> volunteers. All entries being done are from recording (bid) sheets.

It does not matter who actually records the fact that the item has been
sold. The key to the design I have proposed is that this
person NOT be any of the people entering bids. The other key aspect is
that the moment the item record's timestamp is set is the actual moment
when the item has been sold, so if the auctioneer says "Sold!" whoever
is recording that event needs to do it immediately.

> The corner case needed to be handled is this. If two data entry people
> for some reason have recording sheets that has the same catalog
> number, different bid numbers, they enter the bid information and hit
> the <ENTER> key at the same time, both bids get posted. This is what
> we are trying to prevent. If this process is done, say a split second
> apart, The second data entry person is going to be told that the item
> is sold. (If the wrong person got the item then obviously they would
> need to correct the entry error. That is expected in any data entry
> environment.)

I don't see any new information here. However, I suspect we are talking
about different types of auctions. I'm envisioning a computer-mediated
type of auction. Perhaps you are describing a manual auction where the
record keeping is just being done using computers. If so, then I fail to
see the need for having the computer arbitrate whoever won the auction
at all (since the auctioneer and her staff can actually point to the
final buyer and record that person's identification manually.)

I'm convinced my design will fit the needs of the problem, as stated. If
you are not convinced, how about asking some specific questions about
what I have proposed to flush out any hidden problems? Otherwise I
suspect we might stew around in the mud for awhile.