Re: Check volatile functions in ppi_clauses for memoize node

David Rowley <dgrowleyml@gmail.com>

From: David Rowley <dgrowleyml@gmail.com>
To: Richard Guo <guofenglinux@gmail.com>
Cc: PostgreSQL-development <pgsql-hackers@postgresql.org>
Date: 2023-08-07T10:19:08Z
Lists: pgsql-hackers
On Fri, 4 Aug 2023 at 22:26, Richard Guo <guofenglinux@gmail.com> wrote:
> explain (costs off)
> select * from t t1 left join lateral
>     (select t1.a as t1a, t2.a as t2a from t t2) s
> on t1.a = s.t2a + random();
>                   QUERY PLAN
> -----------------------------------------------
>  Nested Loop Left Join
>    ->  Seq Scan on t t1
>    ->  Memoize
>          Cache Key: t1.a
>          Cache Mode: binary
>          ->  Seq Scan on t t2
>                Filter: (t1.a = (a + random()))
> (7 rows)
>
> According to the theory we should not use memoize node for this query
> because of the volatile function in the inner side.  So propose a patch
> to fix that.

Thanks for the patch.  I've pushed a variation of it.

I didn't really see the need to make a single list with all the
RestrictInfos. It seems fine just to write code and loop over the
ppi_clauses checking for volatility.

David



Commits

  1. Don't Memoize lateral joins with volatile join conditions