Revert "ICG updates for ORCA commit: "Include binary coercible casts for predicate inference""

This reverts commit dbece3da.

Performance regression observed for tpcds queries due to cardinality
misestimation. Impacted TPCDS queries 174, 111 and 104.
Signed-off-by: NSambitesh Dash <sdash@pivotal.io>

config/orca.m4

@@ -40,10 +40,10 @@ AC_RUN_IFELSE([AC_LANG_PROGRAM([[
 #include <string.h>
 ]],
 [
-return strncmp("3.18.", GPORCA_VERSION_STRING, 5);
+return strncmp("3.19.", GPORCA_VERSION_STRING, 5);
 ])],
 [AC_MSG_RESULT([[ok]])],
-[AC_MSG_ERROR([Your ORCA version is expected to be 3.18.XXX])]
+[AC_MSG_ERROR([Your ORCA version is expected to be 3.19.XXX])]
 )
 AC_LANG_POP([C++])
 ])# PGAC_CHECK_ORCA_VERSION

configure

@@ -13927,7 +13927,7 @@ int
 main ()
 {
 
-return strncmp("3.18.", GPORCA_VERSION_STRING, 5);
+return strncmp("3.19.", GPORCA_VERSION_STRING, 5);
 
   ;
   return 0;
@@ -13937,7 +13937,7 @@ if ac_fn_cxx_try_run "$LINENO"; then :
   { $as_echo "$as_me:${as_lineno-$LINENO}: result: ok" >&5
 $as_echo "ok" >&6; }
 else
-  as_fn_error $? "Your ORCA version is expected to be 3.18.XXX" "$LINENO" 5
+  as_fn_error $? "Your ORCA version is expected to be 3.19.XXX" "$LINENO" 5
 
 fi
 rm -f core *.core core.conftest.* gmon.out bb.out conftest$ac_exeext \

depends/conanfile_orca.txt

 [requires]
-orca/v3.18.0@gpdb/stable
+orca/v3.19.0@gpdb/stable
 
 [imports]
 include, * -> build/include

gpAux/releng/releng.mk

@@ -109,7 +109,7 @@ sync_tools: opt_write_test
 	-Divyrepo.user=$(IVYREPO_USER) -Divyrepo.passwd="$(IVYREPO_PASSWD)" -quiet resolve);
 
 ifeq "$(findstring aix,$(BLD_ARCH))" ""
-	LD_LIBRARY_PATH='' wget --no-check-certificate -q -O - https://github.com/greenplum-db/gporca/releases/download/v3.18.0/bin_orca_centos5_release.tar.gz | tar zxf - -C $(BLD_TOP)/ext/$(BLD_ARCH)
+	LD_LIBRARY_PATH='' wget --no-check-certificate -q -O - https://github.com/greenplum-db/gporca/releases/download/v3.19.0/bin_orca_centos5_release.tar.gz | tar zxf - -C $(BLD_TOP)/ext/$(BLD_ARCH)
 endif
 
 clean_tools: opt_write_test

src/test/regress/expected/bfv_statistic_optimizer.out

@@ -495,27 +495,26 @@ ANALYZE test_join_card1;
 ANALYZE test_join_card2;
 ANALYZE test_join_card3;
 EXPLAIN SELECT * FROM test_join_card1 t1, test_join_card2 t2, test_join_card3 t3 WHERE t1.b = t2.b and t3.b = t2.b;
-                                                                QUERY PLAN                                                                
-------------------------------------------------------------------------------------------------------------------------------------------
- Gather Motion 3:1  (slice4; segments: 3)  (cost=0.00..1298.38 rows=5999 width=22)
-   ->  Hash Join  (cost=0.00..1297.89 rows=2000 width=22)
-         Hash Cond: (((test_join_card1.b)::text = (test_join_card2.b)::text) AND ((test_join_card3.b)::text = (test_join_card2.b)::text))
-         ->  Hash Join  (cost=0.00..865.03 rows=3334 width=18)
-               Hash Cond: ((test_join_card1.b)::text = (test_join_card3.b)::text)
+                                                  QUERY PLAN                                                   
+---------------------------------------------------------------------------------------------------------------
+ Gather Motion 3:1  (slice4; segments: 3)  (cost=0.00..1297.58 rows=5999 width=22)
+   ->  Hash Join  (cost=0.00..1297.09 rows=2000 width=22)
+         Hash Cond: ((test_join_card2.b)::text = (test_join_card3.b)::text)
+         ->  Hash Join  (cost=0.00..864.43 rows=2000 width=14)
+               Hash Cond: ((test_join_card1.b)::text = (test_join_card2.b)::text)
                ->  Redistribute Motion 3:3  (slice1; segments: 3)  (cost=0.00..431.48 rows=6667 width=10)
-                     Hash Key: test_join_card1.b
+                     Hash Key: (test_join_card1.b)::text
                      ->  Seq Scan on test_join_card1  (cost=0.00..431.15 rows=6667 width=10)
-               ->  Hash  (cost=431.20..431.20 rows=3334 width=8)
-                     ->  Redistribute Motion 3:3  (slice2; segments: 3)  (cost=0.00..431.20 rows=3334 width=8)
-                           Hash Key: (test_join_card3.b)::text
-                           ->  Seq Scan on test_join_card3  (cost=0.00..431.07 rows=3334 width=8)
-         ->  Hash  (cost=431.08..431.08 rows=2000 width=4)
-               ->  Redistribute Motion 3:3  (slice3; segments: 3)  (cost=0.00..431.08 rows=2000 width=4)
-                     Hash Key: (test_join_card2.b)::text
-                     ->  Seq Scan on test_join_card2  (cost=0.00..431.04 rows=2000 width=4)
- Planning time: 596.467 ms
- Optimizer: PQO version 3.9.0
-(18 rows)
+               ->  Hash  (cost=431.08..431.08 rows=2000 width=4)
+                     ->  Redistribute Motion 3:3  (slice2; segments: 3)  (cost=0.00..431.08 rows=2000 width=4)
+                           Hash Key: test_join_card2.b
+                           ->  Seq Scan on test_join_card2  (cost=0.00..431.04 rows=2000 width=4)
+         ->  Hash  (cost=431.20..431.20 rows=3334 width=8)
+               ->  Redistribute Motion 3:3  (slice3; segments: 3)  (cost=0.00..431.20 rows=3334 width=8)
+                     Hash Key: (test_join_card3.b)::text
+                     ->  Seq Scan on test_join_card3  (cost=0.00..431.07 rows=3334 width=8)
+ Optimizer: PQO version 3.19.0
+(17 rows)
 
 -- start_ignore
 DROP TABLE IF EXISTS test_join_card1;

src/test/regress/expected/gporca.out

@@ -11010,96 +11010,3 @@ WITH abc AS (SELECT onetimefilter1.a, onetimefilter1.b FROM onetimefilter1, onet
         1 |        0 | 10
 (10 rows)
 
--- Test predicate inference for certain cast exprs
-create table foid (a oid);
-NOTICE:  Table doesn't have 'DISTRIBUTED BY' clause -- Using column named 'a' as the Greenplum Database data distribution key for this table.
-HINT:  The 'DISTRIBUTED BY' clause determines the distribution of data. Make sure column(s) chosen are the optimal data distribution key to minimize skew.
-create table infer_vc (a varchar);
-NOTICE:  Table doesn't have 'DISTRIBUTED BY' clause -- Using column named 'a' as the Greenplum Database data distribution key for this table.
-HINT:  The 'DISTRIBUTED BY' clause determines the distribution of data. Make sure column(s) chosen are the optimal data distribution key to minimize skew.
-create table infer_txt (a text);
-NOTICE:  Table doesn't have 'DISTRIBUTED BY' clause -- Using column named 'a' as the Greenplum Database data distribution key for this table.
-HINT:  The 'DISTRIBUTED BY' clause determines the distribution of data. Make sure column(s) chosen are the optimal data distribution key to minimize skew.
-CREATE TABLE infer_part_vc (id int, gender varchar(1)) 
-  DISTRIBUTED BY (id)
-  PARTITION BY LIST (gender)
-  ( PARTITION girls VALUES ('F'), 
-    PARTITION boys VALUES ('M'), 
-    DEFAULT PARTITION other );
-NOTICE:  CREATE TABLE will create partition "infer_part_vc_1_prt_girls" for table "infer_part_vc"
-NOTICE:  CREATE TABLE will create partition "infer_part_vc_1_prt_boys" for table "infer_part_vc"
-NOTICE:  CREATE TABLE will create partition "infer_part_vc_1_prt_other" for table "infer_part_vc"
-analyze foid;
-analyze infer_txt;
-analyze infer_vc;
-analyze infer_part_vc;
-explain select * from foid f1 inner join foid f2 on (f1.a = f2.a) where f2.a = 5;
-                                           QUERY PLAN                                           
-------------------------------------------------------------------------------------------------
- Gather Motion 1:1  (slice1; segments: 1)  (cost=10000000000.00..10000000002.01 rows=4 width=8)
-   ->  Nested Loop  (cost=10000000000.00..10000000002.01 rows=2 width=8)
-         ->  Seq Scan on foid f1  (cost=0.00..1.00 rows=1 width=4)
-               Filter: (a = 5::oid)
-         ->  Seq Scan on foid f2  (cost=0.00..1.00 rows=1 width=4)
-               Filter: (a = 5::oid)
- Planning time: 0.776 ms
- Optimizer: legacy query optimizer
-(8 rows)
-
-explain select * from infer_txt f1 inner join infer_txt f2 on (f1.a = f2.a) where f1.a = 'K';
-                                           QUERY PLAN                                            
--------------------------------------------------------------------------------------------------
- Gather Motion 1:1  (slice1; segments: 1)  (cost=10000000000.00..10000000002.01 rows=4 width=64)
-   ->  Nested Loop  (cost=10000000000.00..10000000002.01 rows=2 width=64)
-         ->  Seq Scan on infer_txt f1  (cost=0.00..1.00 rows=1 width=32)
-               Filter: (a = 'K'::text)
-         ->  Seq Scan on infer_txt f2  (cost=0.00..1.00 rows=1 width=32)
-               Filter: (a = 'K'::text)
- Planning time: 0.577 ms
- Optimizer: legacy query optimizer
-(8 rows)
-
-explain select * from infer_vc f1 inner join infer_txt f2 on (f1.a = f2.a) where f1.a = 'K';
-                                           QUERY PLAN                                            
--------------------------------------------------------------------------------------------------
- Gather Motion 1:1  (slice1; segments: 1)  (cost=10000000000.00..10000000002.01 rows=4 width=64)
-   ->  Nested Loop  (cost=10000000000.00..10000000002.01 rows=2 width=64)
-         ->  Seq Scan on infer_vc f1  (cost=0.00..1.00 rows=1 width=32)
-               Filter: ((a)::text = 'K'::text)
-         ->  Seq Scan on infer_txt f2  (cost=0.00..1.00 rows=1 width=32)
-               Filter: (a = 'K'::text)
- Planning time: 0.622 ms
- Optimizer: legacy query optimizer
-(8 rows)
-
-explain select * from infer_vc f1 inner join infer_vc f2 on (f1.a = f2.a) where f1.a = 'K';
-                                           QUERY PLAN                                            
--------------------------------------------------------------------------------------------------
- Gather Motion 1:1  (slice1; segments: 1)  (cost=10000000000.00..10000000002.01 rows=4 width=64)
-   ->  Nested Loop  (cost=10000000000.00..10000000002.01 rows=2 width=64)
-         ->  Seq Scan on infer_vc f1  (cost=0.00..1.00 rows=1 width=32)
-               Filter: ((a)::text = 'K'::text)
-         ->  Seq Scan on infer_vc f2  (cost=0.00..1.00 rows=1 width=32)
-               Filter: ((a)::text = 'K'::text)
- Planning time: 0.524 ms
- Optimizer: legacy query optimizer
-(8 rows)
-
-explain select * from infer_part_vc inner join infer_txt on (infer_part_vc.gender = infer_txt.a) and infer_txt.a = 'M';
-                                            QUERY PLAN                                            
---------------------------------------------------------------------------------------------------
- Gather Motion 3:1  (slice2; segments: 3)  (cost=10000000000.00..10000000004.06 rows=4 width=44)
-   ->  Nested Loop  (cost=10000000000.00..10000000004.06 rows=2 width=44)
-         ->  Redistribute Motion 3:3  (slice1; segments: 3)  (cost=0.00..2.04 rows=1 width=12)
-               Hash Key: infer_part_vc_1_prt_boys.gender
-               ->  Append  (cost=0.00..2.00 rows=1 width=12)
-                     ->  Seq Scan on infer_part_vc_1_prt_boys  (cost=0.00..1.00 rows=1 width=12)
-                           Filter: ((gender)::text = 'M'::text)
-                     ->  Seq Scan on infer_part_vc_1_prt_other  (cost=0.00..1.00 rows=1 width=12)
-                           Filter: ((gender)::text = 'M'::text)
-         ->  Seq Scan on infer_txt  (cost=0.00..1.00 rows=1 width=32)
-               Filter: (a = 'M'::text)
- Planning time: 2.212 ms
- Optimizer: legacy query optimizer
-(13 rows)
-

src/test/regress/sql/gporca.sql

@@ -2047,28 +2047,6 @@ ANALYZE onetimefilter2;
 EXPLAIN WITH abc AS (SELECT onetimefilter1.a, onetimefilter1.b FROM onetimefilter1, onetimefilter2 WHERE onetimefilter1.a=onetimefilter2.a) SELECT (SELECT 1 FROM abc WHERE f1.b = f2.b LIMIT 1), COALESCE((SELECT 2 FROM abc WHERE f1.a=random() AND f1.a=2), 0), (SELECT b FROM abc WHERE b=f1.b) FROM onetimefilter1 f1, onetimefilter2 f2 WHERE f1.b = f2.b;
 WITH abc AS (SELECT onetimefilter1.a, onetimefilter1.b FROM onetimefilter1, onetimefilter2 WHERE onetimefilter1.a=onetimefilter2.a) SELECT (SELECT 1 FROM abc WHERE f1.b = f2.b LIMIT 1), COALESCE((SELECT 2 FROM abc WHERE f1.a=random() AND f1.a=2), 0), (SELECT b FROM abc WHERE b=f1.b) FROM onetimefilter1 f1, onetimefilter2 f2 WHERE f1.b = f2.b;
 
--- Test predicate inference for certain cast exprs
-create table foid (a oid);
-create table infer_vc (a varchar);
-create table infer_txt (a text);
-CREATE TABLE infer_part_vc (id int, gender varchar(1)) 
-  DISTRIBUTED BY (id)
-  PARTITION BY LIST (gender)
-  ( PARTITION girls VALUES ('F'), 
-    PARTITION boys VALUES ('M'), 
-    DEFAULT PARTITION other );
-
-analyze foid;
-analyze infer_txt;
-analyze infer_vc;
-analyze infer_part_vc;
-
-explain select * from foid f1 inner join foid f2 on (f1.a = f2.a) where f2.a = 5;
-explain select * from infer_txt f1 inner join infer_txt f2 on (f1.a = f2.a) where f1.a = 'K';
-explain select * from infer_vc f1 inner join infer_txt f2 on (f1.a = f2.a) where f1.a = 'K';
-explain select * from infer_vc f1 inner join infer_vc f2 on (f1.a = f2.a) where f1.a = 'K';
-explain select * from infer_part_vc inner join infer_txt on (infer_part_vc.gender = infer_txt.a) and infer_txt.a = 'M';
-
 -- start_ignore
 DROP SCHEMA orca CASCADE;
 -- end_ignore