Hi Helen

Thanks for your help. cast('now' as time) works as expected, so that
bug must never have been fixed. It is also still present in the
FB1.5 Beta.

Rgds
Paul

--- In ib-support@yahoogroups.com, Helen Borrie <helebor@t...> wrote:
> At 12:47 AM 7/02/2003 +0000, you wrote:
> >Hi all
> >
> >Using FB1.02
> >
> >I have a group of stored procedures which perform a rather lengthy
> >sequence of calculations and subsequent table inserts. These
> >operations
> >are logged at various intervals in a separate table.
> >
> >The problem I'm having is that the date and time logged for each
> >entry
> >(which is assigned using current_date and current_time) is the
same
> >for
> >every log entry with a single calculation run, even though the run
> >takes
> >a few minutes to complete. I am therefore assuming that the date
and
> >time are assigned when the transactions are finally commited. Is
> >that
> >correct? If so, is there some way I can get the transacations
> >commited
> >at nominated points instead of just when the operation finally
> >completes? It seems that 'commit' is not a valid keyword inside
> >stored
> >procedures.
>
> They *should* be the timestamp when the individual record is
> posted. However, if the logic of your program brings the
timestamp into a
> variable and then uses the variable to post a value when iterating
through
> the loop, of course the value is "frozen".
>
> So, if your proc does
>
> update aTable
> set aTimestamp = CURRENT_TIMESTAMP,
> ....
>
> or if you have a timestamping BeforeUpdate trigger, you should get
the
> timestamp when the row was *posted*, regardless of when the COMMIT
> happens. And, no, transactions are started and completed by the
client,
> not the server, so you can't commit the transaction from inside
itself.
>
> If your logic isn't unintentionally freezing the timestamp,
experiment with
> using CAST ('NOW' AS TIMESTAMP) to see whether it makes a
difference. I
> reported a bug waaaaay back, ex an early beta, regarding the
apparent
> difference between them, and it got deleted from the bug-tracker,
not sure
> whether it was because it was fixed or because the person who
deleted it
> couldn't reproduce it...
>
> heLen