From: | Mike Rylander <mrylander(at)gmail(dot)com> |
---|---|
To: | Christopher Kings-Lynne <chriskl(at)familyhealth(dot)com(dot)au>, pgsql-hackers(at)postgresql(dot)org |
Subject: | Re: Odd plpgsql behaviour |
Date: | 2004-11-15 11:36:29 |
Message-ID: | b918cf3d041115033663f383db@mail.gmail.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
On Mon, 15 Nov 2004 15:12:24 +0800, Christopher Kings-Lynne
<chriskl(at)familyhealth(dot)com(dot)au> wrote:
> On 7.4:
>
> This is what we wanted to do:
>
> IF TG_OP = 'INSERT' OR (TG_OP = 'UPDATE' AND NEW.name != OLD.name) THEN
> EXECUTE x;
> END IF;
>
> However, we had to write it like this:
>
> IF TG_OP = 'INSERT' THEN
> EXECUTE x;
> ELSIF TG_OP = 'UPDATE' AND NEW.name != OLD.name THEN
> EXECUTE x;
> END IF;
>
> Because in the first case it would complain that OLD.name wasn't
> defined, if the trigger was NOT an update.
>
> OK, but the second case works??!?! Is this a weird peculiarity of the
> pl/pgsql lazy evaluation rules? Why doesn't the first one work if the
> second one does?
IIRC, the reason for this is that the entire IF test is passed to the
SQL engine as a SELECT statement after replacing the TG_* identifiers
with their respective values.
Your first example is essentially
IF (SELECT (TG_OP = 'INSERT' OR (TG_OP = 'UPDATE' AND NEW.name !=
OLD.name) IS TRUE) ...
In this case, since OLD.name does not exist during INSERT it cannot be
replaced. Perhaps someone else can shed a little more light on this.
--
Mike Rylander
mrylander(at)gmail(dot)com
GPLS -- PINES Development
Database Developer
From | Date | Subject | |
---|---|---|---|
Next Message | Andrew Dunstan | 2004-11-15 12:35:16 | Re: Concern about new PL/Perl |
Previous Message | Zeugswetter Andreas DAZ SD | 2004-11-15 08:46:11 | Re: psql \e broken again |