Обсуждение: pl/pgsql FOUND variables and RULES

I have a volatile function (trigger) that gathers NEW.*, parses them, then inserts a
subset of values into a different table.

The table is set up as an inherited table where the parent table has a RULE :

CREATE RULE myrule AS ON INSERT TO mytable WHERE ( myfield >= 100 and myfield < 200)
DO INSTEAD
INSERT INTO mytable_100 (field1,field2,field3,myfield) VALUES
(NEW.field1,NEW.field2, NEW.field3, NEW.myfield);

Now when the trigger fires the insert does happen and ends up in mytable_100 but the
FOUND variable is not set (since technically 0 rows got inserted into mytable). Is
there a proper way to set up rules such that the FOUND variable is set within
pl/pgsql upon a successful insert (via the rule) or how can I check for a successful
insert?

Sven

Sven W <sven@dmv.com> writes:
> I have a volatile function (trigger) that gathers NEW.*, parses them, then inserts a
> subset of values into a different table.

> The table is set up as an inherited table where the parent table has a RULE :

> CREATE RULE myrule AS ON INSERT TO mytable WHERE ( myfield >= 100 and myfield < 200)
> DO INSTEAD
> INSERT INTO mytable_100 (field1,field2,field3,myfield) VALUES
> (NEW.field1,NEW.field2, NEW.field3, NEW.myfield);

> Now when the trigger fires the insert does happen and ends up in mytable_100 but the
> FOUND variable is not set (since technically 0 rows got inserted into mytable). Is
> there a proper way to set up rules such that the FOUND variable is set within
> pl/pgsql upon a successful insert (via the rule) or how can I check for a successful
> insert?

On the whole, you'd be a lot better off letting the trigger function
handle the redirection into the proper child table, I'd think.  Using
both a trigger and a rule seems like it mainly gives you the worst of
both worlds.

            regards, tom lane