Return-Path: pg_adm@postgres.berkeley.edu
Received: by postgres.Berkeley.EDU (5.61/1.29)
	id AA03623; Mon, 22 Mar 93 01:59:16 -0800
Date: Mon, 22 Mar 93 01:59:16 -0800
Message-Id: <9303220959.AA03623@postgres.Berkeley.EDU>
From: mcquaig!postgres@uunet.UU.NET (Postgres System User)
Subject: NULL problems are still around
To: postgres@postgres.berkeley.edu
Sender: pg_adm@postgres.berkeley.edu


Strange things happen when a rule tries to access a datum that ISNULL
as in the following:

define rule pm_FOBInit is 
on append to poorme
do [
	replace new (FOB = mystuff.FOB) where mystuff.key = new.key
] \g

If the rule is triggered and FOB ISNULL you get an invalid pointer
error:  WARN:Mar 22 01:34:50:datumGetSize: Invalid Datum Pointer
Things then tend to get strange and core dumps, even kernel panics
have been know to happen shortly afterward.  Three kernel panics
before I found it.

BTW FOB is of type text.  I thought perhaps type casting might be a
culprit but postgres knew that it was varlena.  I guess something
upstream actually tries to use the data (ie. access the len and copy!)
which could well send it to never never land.  NULL problems it seems
are still with us.
nmm

Neil M. McQuaig      344 Millicent Way,  Shreveport, LA  71106
VOICE: (318)868-5611 UUCP: mcquaig!nmm (318)861-1051 or uunet!mcquaig!nmm