From 219f3676a07e2e69e0714155791c7da8495532a9 Mon Sep 17 00:00:00 2001
From: Onder Kalaci <onder@citusdata.com>
Date: Tue, 24 Sep 2019 11:51:05 +0200
Subject: [PATCH] Improve some tests around local execution and CTE inlining on
 pg 12

---
 .../expected/local_shard_execution.out        | 19 ++++++++++++++++++-
 .../expected/local_shard_execution_0.out      | 19 ++++++++++++++++++-
 .../regress/sql/local_shard_execution.sql     | 14 +++++++++++++-
 3 files changed, 49 insertions(+), 3 deletions(-)

diff --git a/src/test/regress/expected/local_shard_execution.out b/src/test/regress/expected/local_shard_execution.out
index 7f3bbb3ee..e97c6dde5 100644
--- a/src/test/regress/expected/local_shard_execution.out
+++ b/src/test/regress/expected/local_shard_execution.out
@@ -699,6 +699,9 @@ ORDER BY
 
 -- multi-shard CTE is followed by a query which could be executed locally, but
 -- since the query started with a parallel query, it doesn't use local execution
+-- note that if we allow Postgres to inline the CTE (e.g., not have the EXISTS 
+-- subquery), then it'd pushdown the filters and the query becomes single-shard, 
+-- locally executable query
 WITH all_data AS (SELECT * FROM distributed_table)
 SELECT 
 	count(*) 
@@ -706,13 +709,27 @@ FROM
 	distributed_table, all_data 
 WHERE 
 	distributed_table.key = all_data.key AND distributed_table.key = 1
-	-- the following is to avoid CTE inlining
 	AND EXISTS (SELECT * FROM all_data);
  count 
 -------
      1
 (1 row)
 
+-- in pg12, the following CTE can be inlined, still the query becomes
+-- a subquery that needs to be recursively planned and a parallel
+-- query, so do not use local execution
+WITH all_data AS (SELECT age FROM distributed_table)
+SELECT 
+	count(*) 
+FROM 
+	distributed_table, all_data 
+WHERE 
+	distributed_table.key = all_data.age AND distributed_table.key = 1;
+ count 
+-------
+     0
+(1 row)
+
 -- get ready for the next commands
 TRUNCATE reference_table, distributed_table, second_distributed_table;
 -- local execution of returning of reference tables
diff --git a/src/test/regress/expected/local_shard_execution_0.out b/src/test/regress/expected/local_shard_execution_0.out
index b37e37a9a..2b67c40fc 100644
--- a/src/test/regress/expected/local_shard_execution_0.out
+++ b/src/test/regress/expected/local_shard_execution_0.out
@@ -685,6 +685,9 @@ ORDER BY
 
 -- multi-shard CTE is followed by a query which could be executed locally, but
 -- since the query started with a parallel query, it doesn't use local execution
+-- note that if we allow Postgres to inline the CTE (e.g., not have the EXISTS 
+-- subquery), then it'd pushdown the filters and the query becomes single-shard, 
+-- locally executable query
 WITH all_data AS (SELECT * FROM distributed_table)
 SELECT 
 	count(*) 
@@ -692,13 +695,27 @@ FROM
 	distributed_table, all_data 
 WHERE 
 	distributed_table.key = all_data.key AND distributed_table.key = 1
-	-- the following is to avoid CTE inlining
 	AND EXISTS (SELECT * FROM all_data);
  count 
 -------
      1
 (1 row)
 
+-- in pg12, the following CTE can be inlined, still the query becomes
+-- a subquery that needs to be recursively planned and a parallel
+-- query, so do not use local execution
+WITH all_data AS (SELECT age FROM distributed_table)
+SELECT 
+	count(*) 
+FROM 
+	distributed_table, all_data 
+WHERE 
+	distributed_table.key = all_data.age AND distributed_table.key = 1;
+ count 
+-------
+     0
+(1 row)
+
 -- get ready for the next commands
 TRUNCATE reference_table, distributed_table, second_distributed_table;
 -- local execution of returning of reference tables
diff --git a/src/test/regress/sql/local_shard_execution.sql b/src/test/regress/sql/local_shard_execution.sql
index 7cd7e4027..1aea11abc 100644
--- a/src/test/regress/sql/local_shard_execution.sql
+++ b/src/test/regress/sql/local_shard_execution.sql
@@ -419,6 +419,9 @@ ORDER BY
 
 -- multi-shard CTE is followed by a query which could be executed locally, but
 -- since the query started with a parallel query, it doesn't use local execution
+-- note that if we allow Postgres to inline the CTE (e.g., not have the EXISTS 
+-- subquery), then it'd pushdown the filters and the query becomes single-shard, 
+-- locally executable query
 WITH all_data AS (SELECT * FROM distributed_table)
 SELECT 
 	count(*) 
@@ -426,9 +429,18 @@ FROM
 	distributed_table, all_data 
 WHERE 
 	distributed_table.key = all_data.key AND distributed_table.key = 1
-	-- the following is to avoid CTE inlining
 	AND EXISTS (SELECT * FROM all_data);
 
+-- in pg12, the following CTE can be inlined, still the query becomes
+-- a subquery that needs to be recursively planned and a parallel
+-- query, so do not use local execution
+WITH all_data AS (SELECT age FROM distributed_table)
+SELECT 
+	count(*) 
+FROM 
+	distributed_table, all_data 
+WHERE 
+	distributed_table.key = all_data.age AND distributed_table.key = 1;
 
 -- get ready for the next commands
 TRUNCATE reference_table, distributed_table, second_distributed_table;