Re: Usage of RowId...

> I don't like that idea because it essentially requires two updates to
> achieve one update,
> I think i see the potential for a race condition....
> SQL SERVER/SYBASE have the perfect solution in their timestamp column,
>
> the server itself generates a unique number for the row every time it
> updates,
> therefor if everone is reading the timestamp, and everyone updates
> using the timestamp, it is gaurenteed that ONE update is always
> sucessful!!! - the others fail and need to
> try again.

This isn't a feature. This is a kludge required by the fact that SQL Server (and Sybase up to recently) didn't have record level locking.

Daniel A. Morgan Received on Thu Nov 26 1998 - 00:00:00 CET