From: Tom Lane <tgl@sss.pgh.pa.us>
To: christianmduta@gmail.com
cc: pgsql-bugs@postgresql.org
Subject: Re: BUG #14924: Subquery in VALUES inside recursive CTE
In-reply-to: <20171124120836.1463.5310@wrigleys.postgresql.org>
References: <20171124120836.1463.5310@wrigleys.postgresql.org>
Comments: In-reply-to christianmduta@gmail.com
	message dated "Fri, 24 Nov 2017 12:08:36 +0000"
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <29656.1511543186.1@sss.pgh.pa.us>
Date: Fri, 24 Nov 2017 12:06:26 -0500
Message-ID: <29657.1511543186@sss.pgh.pa.us>
Precedence: bulk

christianmduta@gmail.com writes:
> When working with recursive CTEs, I had the following happen:

Thanks for the report!  Seems not to be specific to CTEs:

regression=# SELECT val.x
  FROM generate_series(1,10) AS s(i),
  LATERAL (
    VALUES
    ((SELECT s.i + 1)), 
    (s.i + 1)
  ) AS val(x)
  WHERE s.i < 10;
ERROR:  SubPlan found with no parent plan

However, simplifying further, this works:

regression=# SELECT val.x
  FROM generate_series(1,10) AS s(i),
  LATERAL (
    VALUES
    ((SELECT 1 + 1)), 
    (s.i + 1)
  ) AS val(x)
  WHERE s.i < 10;

so it looks like it's the LATERAL reference inside the sub-select
that is needed to trigger the bug.

Same results back to 9.3 :-(

			regards, tom lane