Re: PL/pgSQL return value in after triggers - Mailing list pgsql-hackers

From Pavel Stehule
Subject Re: PL/pgSQL return value in after triggers
Date
Msg-id CAFj8pRARVKiLMztaCZWcZ7Jn0qGTKG5G4s97dHGLz=ROHsbcTQ@mail.gmail.com
Whole thread Raw
In response to Re: PL/pgSQL return value in after triggers  (Peter Eisentraut <peter_e@gmx.net>)
List pgsql-hackers
Hello Peter

I checked code, and I don't think so this is good.

A design of optional NULL is going to inconsistent syntax.

RETURN (OLD, NEW, NULL, /* nothing */) is not consistent

But my main argument is not intuitive behave of BEFORE triggers after
this change.

When somebody write  BEFORE trigger function like:

BEGIN RAISE NOTICE '%', NEW.x; RETURN;
END;

then don't expect so all rows will be lost.

Preferred default return value for BEFORE INSERT UPDATE trigger should
be NEW, and for DELETE trigger should be OLD - not NULL.

And because we cannot to distinct between BEFORE and AFTER trigger in
parser, I propose don't change current behave. Current behave is not
too friendly - but is consistent with simple rules.

Regards

Pavel


2012/1/2 Peter Eisentraut <peter_e@gmx.net>:
> On mån, 2011-02-28 at 19:07 +0200, Peter Eisentraut wrote:
>> PL/pgSQL trigger functions currently require a value to be returned,
>> even though that value is not used for anything in case of a trigger
>> fired AFTER.  I was wondering if we could relax that.  It would make
>> things a bit more robust and produce clearer PL/pgSQL code.  The
>> specific case I'm concerned about is that a trigger function could
>> accidentally be run in a BEFORE trigger even though it was not meant for
>> that.  It is common practice that trigger functions for AFTER triggers
>> return NULL, which would have unpleasant effects if used in a BEFORE
>> trigger.
>>
>> I think it is very uncommon to have the same function usable for BEFORE
>> and AFTER triggers, so it would be valuable to have coding support
>> specifically for AFTER triggers.  We could just allow RETURN without
>> argument, or perhaps no RETURN at all.
>
> Here is a patch for that.
>
> One thing that I'm concerned about with this is that it treats a plain
> RETURN in a BEFORE trigger as RETURN NULL, whereas arguably it should be
> an error.  I haven't found a good way to handle that yet, but I'll keep
> looking.
>
>
>
> --
> Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
> To make changes to your subscription:
> http://www.postgresql.org/mailpref/pgsql-hackers
>


pgsql-hackers by date:

Previous
From: Merlin Moncure
Date:
Subject: Re: GUC_REPORT for protocol tunables was: Re: Optimize binary serialization format of arrays with fixed size elements
Next
From: Merlin Moncure
Date:
Subject: Re: Re: Add minor version to v3 protocol to allow changes without breaking backwards compatibility