Stan wrote:

>
> According to The great Firebird Book, try/catch blocks
> within a stored procedure act as "nested" transactions.
>

The nested transactions have the same snapshot of the database
as the outer transaction, so asking the question over and over
is going to get the same answer for snapshot/concurrency
transactions. If you run that procedure in read-committed
mode, it will eventually see the conflicting record and the
new data when the second transaction commits. On the other
hand, it will waste a lot of CPU testing over and over until
the commit happens.

So generally, I say no, that's not a good way to hand lock
conflicts.


Regards,


Ann