Adam wrote:

> --- In Firebird-Architect@yahoogroups.com, "Ann W. Harrison"
> <aharrison@...> wrote:
>> Adam wrote:
>>> Perhaps both 'insert or update' and 'update or insert' could be both
>>> implemented, and the one you choose would determine which triggers are
>>> fired first.
>>>
>> for what little it's worth, Firebird triggers fire on row update, not
>> on verb, so an insert that fails because of a duplicate key value
>> doesn't file the insert trigger, nor does an update that finds no
>> rows. The question of "which triggers are fired first" doesn't come
>> up.
>
> The before insert trigger(s) would fire regardless, otherwise the
> autoinc simulation
>
> CREATE TRIGGER A_BI FOR A
> ACTIVE BEFORE INSERT POSITION 0
> AS
> BEGIN
> IF (NEW.ID IS NULL) THEN
> BEGIN
> NEW.ID = GEN_ID(GEN_AID, 1);
> END
> END
> ^
>
> would fail to work.
>
> Running an update which affects 0 rows would not fire any update trigger.
>
> So if it tried insert then update:
>
> Before insert trigger -> attempt insert -> fails -> before update
> trigger -> update -> after update trigger.
>
> Or if it tried update and insert:
>
> The update wouldn't affect any row, so no update triggers are fired,
> therefore the insert is attempted.
>
> So depending on the methodology chosen, the before insert triggers may
> be fired.

Good pickup Adam - I knew there was a reason I try the update first ;)

--
Lester Caine - G8HFL
-----------------------------
L.S.Caine Electronic Services - http://home.lsces.co.uk
Model Engineers Digital Workshop -
http://home.lsces.co.uk/ModelEngineersDigitalWorkshop/
Treasurer - Firebird Foundation Inc. - http://www.firebirdsql.org/index.php