Re: triggers vs. 'NEW in non-rule query' -- SOLVED?

will trillich wrote:
>
> i THINK i've got it! MWAAH HA HA...
>

Congrats

> i follow the example from the postgresql-doc/html/user/c4*40.html
> document and define a function to be called via TRIGGER...
>
> CREATE FUNCTION easy() RETURNS OPAQUE AS '
> DECLARE
> x CHAR(1) := NEW.somefield FROM 1 FOR 1;
> BEGIN
> -- whatever, yada yada, bada bing bada boom
> RETURN NEW;
> END;
> ' language 'plpgsql';

Grmpf - exactly that's missing in the "#option dump" output -
the calculation of default values for variable
initialization. I was looking at your log-snippent over and
over again and couldn't see it :-)

>
> and this results in "ERROR: use of NEW in non-rule procedure"
> when actually called via TRIGGER. i change it instead to
>
> CREATE FUNCTION easy() RETURNS OPAQUE AS '
> DECLARE
> x CHAR(1); -- no := assignment here
> BEGIN
> x := NEW.somefield FROM 1 FOR 1;
> -- whatever, yada yada, bada bing bada boom
> RETURN NEW;
> END;
> ' language 'plpgsql';
>
> and suddenly all is right with the world. so i can't refer to the
> NEW (or probably OLD) pseudotables WITHIN THE DECLARE section.
> apparently.
>
> Q: is this a bug? is this a feature?

A missing feature, and something that should be on 7.2's
TODO.

Jan

--

#======================================================================#
# It's easier to get forgiveness for being wrong than for being right. #
# Let's break this rule - forgive me. #
#================================================== JanWieck(at)Yahoo(dot)com #

_________________________________________________________
Do You Yahoo!?
Get your free @yahoo.com address at http://mail.yahoo.com