Re: [HACKERS] SPI_prepare() doesn't work well ?

> -----Original Message-----
> From: owner-pgsql-hackers(at)postgreSQL(dot)org
> [mailto:owner-pgsql-hackers(at)postgreSQL(dot)org]On Behalf Of Jan Wieck
> Sent: Thursday, January 21, 1999 12:57 AM
> To: Hiroshi Inoue
> Cc: pgsql-hackers(at)postgreSQL(dot)org; mcontzen(at)dohle(dot)com
> Subject: Re: [HACKERS] SPI_prepare() doesn't work well ?
>
>
> Hiroshi Inoue wrote:
>
> >
> > Hello all,
> >
> > It seems that SPI_prepare() doesn't work well in some cases.
> >
> > Pawel Pierscionek [pawel(at)astercity(dot)net] reported about the
> > following case 1([SQL] drop table in pgsql).
> > Michael Contzen [mcontzen(at)dohle(dot)com] reported about the
> > following case 2(PL/PGSQL bug using aggregates).
> > > You can find it from pgsql-hackers archive.
> > >
>
> [snip]
>
> > >
> > > 2. Aggregates in PL/pgSQL cause wrong results.
> > >
> > > Is it proper to use set_agg_tlist_references() function to
> > > reconstruct aggs member node for Agg type nodes ?
> >
> > Don't know. It is important, that the copy of the tree has
> > absolutely NO references to anything outside itself. The
> > parser/rewrite/planner combo creates all plans in the actual
> > transactions memory context. So they will get destroyed at
> > transaction end.
> >
>
> In my patch set_agg_tlist_refereces() function is appiled for
> copied(new) Agg type nodes,not for original(old) nodes.
>
> This case is a special case of
>
> How does copyObject() copy multiply referenced objects ?
>
> As to this case,PostgreSQL Executor sets aggno's of Aggreg Nodes
> using aggs member node of Agg type nodes.
> As a result,aggno's of the Aggreg nodes contained in TargetList
> are set because they have common pointers.
> So multiply referenced Aggreg type nodes must be copied to new
> plan as multiply referenced nodes.
>

It is my understanding that this has always been a problem. I have
never 100% been confident I understand it.

As I remember, there used to be a:

Aggreg **qry_agg

that was a member of the Query structure, and cause all sorts of
problems because the Agg was in the target list _and_ in the qry_agg.
That was removed. Looks like I did it, but I don't remember doing it:

revision 1.44
date: 1998/01/15 19:00:11; author: momjian; state: Exp; lines: +2 -4
Remove Query->qry_aggs and qry_numaggs and replace with Query->hasAggs.

Pass List* of Aggregs into executor, and create needed array there.
No longer need to double-processs Aggregs with second copy in Query.

The removal of that fixed many problems. Can you explain a little more
on how multiple Agg references happen. Perhaps we can get a good clean
fix for this. Maybe redesign is needed, as I did for the removal of
qry_aggs.

--
Bruce Momjian | http://www.op.net/~candle
maillist(at)candle(dot)pha(dot)pa(dot)us | (610) 853-3000
+ If your life is a hard drive, | 830 Blythe Avenue
+ Christ can be your backup. | Drexel Hill, Pennsylvania 19026