Re: Optimizing nested ConvertRowtypeExpr execution

Sorry for the absense.

At Sun, 4 Nov 2018 16:26:12 +0100, Dmitry Dolgov <9erthalion6@gmail.com> wrote in
<CA+q6zcWMCNNmMQ-csuDf0Pqr1_ESat5-Vcu5uognfS3EaC4Apg@mail.gmail.com>
> > On Sun, 4 Nov 2018 at 15:48, Andrew Gierth <andrew@tao11.riddles.org.uk> wrote:
> >
> > >>>>> "Dmitry" == Dmitry Dolgov <9erthalion6@gmail.com> writes:
> >
> >  Dmitry> This patch went through the last tree commit fests without any
> >  Dmitry> noticeable activity, but cfbot says it still applies and
> >  Dmitry> doesn't break any tests. The patch itself is rather small, and
> >  Dmitry> I could reproduce ~20% of performance improvements while
> >  Dmitry> running the same scripts under pgbench (although not in all
> >  Dmitry> cases), but probably we need to find someone to take over it.
> >  Dmitry> Does anyone wants to do so, maybe Kyotaro?
> >
> > I'll deal with it.
> 
> Thanks!

My last comment was the while() loop does nothing. Ashutosh said
that it is on the model of RelabelType.

I examined the code for T_RelabelType again and it is intending
that the ece_mutator can convert something (specifically only
CollateExpr) to RelableType, then reduce the nested Relabels. So
the order is not wrong in this perspective.

So I don't object to the current patch, but it needs test like
attached.

regards.

-- 
Kyotaro Horiguchi
NTT Open Source Software Center
From 45ef5b753069eb89ab5139828884fbdbf0958778 Mon Sep 17 00:00:00 2001
From: Kyotaro Horiguchi <horiguchi.kyotaro@lab.ntt.co.jp>
Date: Tue, 6 Nov 2018 17:12:17 +0900
Subject: [PATCH 2/2] Test code for Optimize nested ConvertRowtypExprs

---
 src/test/regress/expected/inherit.out | 18 ++++++++++++++++++
 src/test/regress/sql/inherit.sql      |  5 +++++
 2 files changed, 23 insertions(+)

diff --git a/src/test/regress/expected/inherit.out b/src/test/regress/expected/inherit.out
index 4f29d9f891..1474ed8190 100644
--- a/src/test/regress/expected/inherit.out
+++ b/src/test/regress/expected/inherit.out
@@ -764,6 +764,8 @@ NOTICE:  drop cascades to table c1
 -- tables. See the pgsql-hackers thread beginning Dec. 4/04
 create table base (i integer);
 create table derived () inherits (base);
+create table more_derived (like derived, b int) inherits (derived);
+NOTICE:  merging column "i" with inherited definition
 insert into derived (i) values (0);
 select derived::base from derived;
  derived 
@@ -777,6 +779,22 @@ select NULL::derived::base;
  
 (1 row)
 
+-- remove redundant conversions.
+explain (verbose on, costs off) select row(i, b)::more_derived::derived::base from more_derived;
+                QUERY PLAN                 
+-------------------------------------------
+ Seq Scan on public.more_derived
+   Output: (ROW(i, b)::more_derived)::base
+(2 rows)
+
+explain (verbose on, costs off) select (1, 2)::more_derived::derived::base;
+      QUERY PLAN       
+-----------------------
+ Result
+   Output: '(1)'::base
+(2 rows)
+
+drop table more_derived;
 drop table derived;
 drop table base;
 create table p1(ff1 int);
diff --git a/src/test/regress/sql/inherit.sql b/src/test/regress/sql/inherit.sql
index a6e541d4da..8308330fed 100644
--- a/src/test/regress/sql/inherit.sql
+++ b/src/test/regress/sql/inherit.sql
@@ -237,9 +237,14 @@ drop table p1 cascade;
 -- tables. See the pgsql-hackers thread beginning Dec. 4/04
 create table base (i integer);
 create table derived () inherits (base);
+create table more_derived (like derived, b int) inherits (derived);
 insert into derived (i) values (0);
 select derived::base from derived;
 select NULL::derived::base;
+-- remove redundant conversions.
+explain (verbose on, costs off) select row(i, b)::more_derived::derived::base from more_derived;
+explain (verbose on, costs off) select (1, 2)::more_derived::derived::base;
+drop table more_derived;
 drop table derived;
 drop table base;
 
-- 
2.16.3