This would have the same problem. Unless there is some way of
recording the commit time in all the rows.

Thanks!

--- In firebird-support@yahoogroups.com, Alexandre Benson Smith
<iblist@t...> wrote:

>
> kapsee wrote:
>
> >I have a C application that caches data from some database tables.
> >The tables have a trigger that update the row with a generator id
> >whenever the row changes. Every time the cache is refreshed from
> >tables to memory, the last id read is kept track of. Next time, the
> >cache is refreshed, to avoid reading the entire dataset, only the rows
> >having id greater than the last id are read. The problem with this
> >is that after the cache is refreshed, a transaction with a lower
> >generator id might commit and the caching scheme will miss this row.
> >
> >Maybe this is not a good way to do this. Anyone has other suggestions ?
> >
> >Thanks!
> >
> >
> Use a Timestamp.
>
> see you !
>
> --
> Alexandre Benson Smith
> Development
> THOR Software e Comercial Ltda
> Santo Andre - Sao Paulo - Brazil
> www.thorsoftware.com.br
>