On Tue, 25 Mar 2003, Adrian Pop wrote:
> I have a postgresql 7.3.2 crash.
> Below you have the details.
Hmm, I can reproduce this with current sources. Backtrace from core
looks like a bug in the setdefault trigger (I'm at work so I don't
have real access to source to get more details).
> Now a question: What is with the folowing statement in the log?
> 2003-03-25 18:42:27 [3531] LOG: query: UPDATE ONLY "public"."p_title" SET "name_id" = NULL WHERE "name_id" = $1
> Why is set to NULL when it fact it should be set to default (in this case 0) according to
> the "on delete set default" from p_title definition?
That's because the set default uses a pretty ugly hack. It plans the
query with =NULL and then replaces the NULL with the default value's
stored plan info. My first guess is that it's something with this that's
causing the crash.