From: Tom Lane <tgl@sss.pgh.pa.us>
To: David Rowley <david.rowley@2ndquadrant.com>
cc: Alvaro Herrera <alvherre@2ndquadrant.com>,
        Robert Haas <robertmhaas@gmail.com>,
        Jeevan Chalke <jeevan.chalke@enterprisedb.com>,
        PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>
Subject: Re: Soon-to-be-broken regression test case
In-reply-to: 
 <CAKJS1f_0A42gP94CWduAamThYWdY6=VeUZ0u4RvnOPbDiDWA1w@mail.gmail.com>
References: <5749.1539276324@sss.pgh.pa.us>
 <20181011165251.k7ggwmvmlfjv33gr@alvherre.pgsql>
 <CAKJS1f_0A42gP94CWduAamThYWdY6=VeUZ0u4RvnOPbDiDWA1w@mail.gmail.com>
Comments: In-reply-to David Rowley <david.rowley@2ndquadrant.com>
	message dated "Fri, 12 Oct 2018 08:45:37 +1300"
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <23506.1539290744.1@sss.pgh.pa.us>
Date: Thu, 11 Oct 2018 16:45:44 -0400
Message-ID: <23507.1539290744@sss.pgh.pa.us>
Precedence: bulk

David Rowley <david.rowley@2ndquadrant.com> writes:
> I guess if we ever did something to break that then we'd need to not
> do anything when there are volatile functions present.

Yeah, nothing I'm doing here changes the rule that we don't flatten
sub-selects containing volatiles in their tlist.

> If people are
> writing that then probably they're doing so to trick the planner,
> perhaps to hide some stats that get outdated easily. I'd imagine we'd
> upset more people than we'd please.

The specific case I'm aware of is that people sometimes write
"(SELECT x)" rather than just "x" so as to make the calculation
be a done-only-once InitPlan.  That code path isn't affected by
this, either (and that's why the partition_prune tests didn't
change behavior).

It's fair to wonder whether partition_prune needs to be testing
other subplan cases besides InitPlans, but that seems like a
distinct issue.

			regards, tom lane