Re: Removing [Merge]Append nodes which contain a single subpath

From: Ashutosh Bapat <ashutosh(dot)bapat(at)enterprisedb(dot)com>
To: David Rowley <david(dot)rowley(at)2ndquadrant(dot)com>
Cc: PostgreSQL-development <pgsql-hackers(at)postgresql(dot)org>
Subject: Re: Removing [Merge]Append nodes which contain a single subpath
Date: 2017-10-26 10:12:18
Message-ID: CAFjFpRcKmFWvbY=b7A69LZpFnVEV32F4uuMCaKHQLARefNQF8g@mail.gmail.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

On Thu, Oct 26, 2017 at 3:29 AM, David Rowley
<david(dot)rowley(at)2ndquadrant(dot)com> wrote:
> However, method 2 appears to also require some Var
> translation in Path targetlists which contain this Proxy path, either
> that or some global Var replacement would need to be done during
> setrefs.
>

For inheritance and partitioning, we translate parent expressions to
child expression by applying Var translations. The translated
expressions differ only in the Var nodes (at least for partitioning
this true, and I believe it to be true for inheritance). I have been
toying with an idea of having some kind of a (polymorphic?) Var node
which can represent parent and children. That will greatly reduce the
need to translate the expression trees and will also help in this
implementation. I am wondering, if ProxyPath could be helpful in
avoiding the cost of reparameterize_path_by_child() introduced for
partition-wise join.

--
Best Wishes,
Ashutosh Bapat
EnterpriseDB Corporation
The Postgres Database Company

In response to

Browse pgsql-hackers by date

  From Date Subject
Next Message Andrey Borodin 2017-10-26 10:22:29 Index only scan for cube and seg
Previous Message Kyotaro HORIGUCHI 2017-10-26 10:05:51 Re: [BUGS] Bug in Physical Replication Slots (at least 9.5)?