From: Tom Lane <tgl@sss.pgh.pa.us>
To: Andrew Dunstan <andrew.dunstan@2ndquadrant.com>
cc: pgsql-hackers@postgresql.org
Subject: Re: Domains and arrays and composites, oh my
In-reply-to: <044f3a1c-9e20-c08b-8f78-c47aa41d7a0c@2ndQuadrant.com>
References: <4206.1499798337@sss.pgh.pa.us> <12364.1499973554@sss.pgh.pa.us>
 <aaaa73ea-4d5f-ed8c-7222-06a2d5f2891e@2ndQuadrant.com>
 <32440.1506618167@sss.pgh.pa.us>
 <044f3a1c-9e20-c08b-8f78-c47aa41d7a0c@2ndQuadrant.com>
Comments: In-reply-to Andrew Dunstan <andrew.dunstan@2ndquadrant.com>
	message dated "Thu, 28 Sep 2017 14:58:00 -0400"
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <13130.1506625418.1@sss.pgh.pa.us>
Date: Thu, 28 Sep 2017 15:03:38 -0400
Message-ID: <13131.1506625418@sss.pgh.pa.us>
Precedence: bulk
Sender: pgsql-hackers-owner@postgresql.org

Andrew Dunstan <andrew.dunstan@2ndquadrant.com> writes:
> On 09/28/2017 01:02 PM, Tom Lane wrote:
>>> I do think that treating a function returning a domain-over-composite
>>> differently from one returning a base composite is a POLA. We'd be very
>>> hard put to explain the reasons for it to an end user.

>> Do you have any thoughts about how we ought to resolve that?

> Not offhand. Maybe we need to revisit the decision not to modify the
> executor at all.

I think it's more of a parse analysis change: the issue is whether to
smash a function's result type to base when determining whether it emits
columns.  Maybe we could just do that in that context, and otherwise leave
domains alone.

> One thought I had was that we could invent a new return
> type of TYPEFUNC_DOMAIN_COMPOSITE so there would be less danger of a PL
> just treating it as an unconstrained base type as it might do if it saw
> TYPEFUNC_COMPOSITE.

Hmm.  That would be a way of forcing the issue, no doubt ...

			regards, tom lane


-- 
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers