Fix segfault in continuous aggregate creation on PG18

When a continuous aggregate query had functionally dependant
columns that were not part of the GROUP BY clause the create
statement would segfault.

Author: svenklemm

.unreleased/pr_9303

@@ -0,0 +1 @@
+Fixes: #9303 Fix segfault in continuous aggregate creation on PG18

tsl/src/continuous_aggs/finalize.c

@@ -257,20 +257,27 @@ mattablecolumninfo_addentry(MaterializationHypertableColumnInfo *out, Node *inpu
 				Var *var = castNode(Var, tle->expr);
 				Assert((int) var->varno <= list_length(rtable));
 				RangeTblEntry *rte = list_nth(rtable, var->varno - 1);
-				Assert(rte->rtekind == RTE_GROUP);
 				Assert(var->varattno > 0);
-				Node *node = list_nth(rte->groupexprs, var->varattno - 1);
-				if (IsA(node, FuncExpr))
+				/*
+				 * Var might not be part of the GROUP BY clause
+				 * eg for functionally dependent columns on tables with primary key
+				 */
+				if (rte->rtekind == RTE_GROUP)
 				{
-					if (contain_mutable_functions(node))
+					Node *node = list_nth(rte->groupexprs, var->varattno - 1);
+					if (IsA(node, FuncExpr))
 					{
-						ereport(WARNING,
-								(errmsg("using non-immutable functions in continuous aggregate "
-										"view may lead to "
-										"inconsistent results on rematerialization")));
+						if (contain_mutable_functions(node))
+						{
+							ereport(WARNING,
+									(errmsg("using non-immutable functions in continuous aggregate "
+											"view may lead to "
+											"inconsistent results on rematerialization")));
+						}
+						FuncExpr *expr = (FuncExpr *) node;
+						timebkt_chk =
+							function_allowed_in_cagg_definition(((FuncExpr *) expr)->funcid);
 					}
-					FuncExpr *expr = (FuncExpr *) node;
-					timebkt_chk = function_allowed_in_cagg_definition(((FuncExpr *) expr)->funcid);
 				}
 			}
 #endif

tsl/test/expected/cagg_ddl-15.out

@@ -2181,3 +2181,14 @@ SELECT view_name from timescaledb_information.continuous_aggregates WHERE hypert
 -------------------
  cagg_hypertab_ddl
 
+-- TEST continuous aggregate with functionally dependent columns
+-- name column depends on id which is in GROUP BY so name doesnt have to be
+CREATE table sensor(id int PRIMARY KEY, name text);
+CREATE TABLE sensordata(time timestamptz,id int, value float) WITH	(timescaledb.hypertable);
+NOTICE:  using column "time" as partitioning column
+CREATE MATERIALIZED VIEW cagg_sensordata WITH (tsdb.continuous) AS
+SELECT s.id, s.name,time_bucket('15 minutes', sd.time) as bucket, avg(sd.value) FROM sensordata sd JOIN sensor s USING(id) GROUP BY s.id, bucket WITH NO DATA;
+SELECT * FROM cagg_sensordata;
+ id | name | bucket | avg 
+----+------+--------+-----
+

tsl/test/expected/cagg_ddl-16.out

@@ -2181,3 +2181,14 @@ SELECT view_name from timescaledb_information.continuous_aggregates WHERE hypert
 -------------------
  cagg_hypertab_ddl
 
+-- TEST continuous aggregate with functionally dependent columns
+-- name column depends on id which is in GROUP BY so name doesnt have to be
+CREATE table sensor(id int PRIMARY KEY, name text);
+CREATE TABLE sensordata(time timestamptz,id int, value float) WITH	(timescaledb.hypertable);
+NOTICE:  using column "time" as partitioning column
+CREATE MATERIALIZED VIEW cagg_sensordata WITH (tsdb.continuous) AS
+SELECT s.id, s.name,time_bucket('15 minutes', sd.time) as bucket, avg(sd.value) FROM sensordata sd JOIN sensor s USING(id) GROUP BY s.id, bucket WITH NO DATA;
+SELECT * FROM cagg_sensordata;
+ id | name | bucket | avg 
+----+------+--------+-----
+

tsl/test/expected/cagg_ddl-17.out

@@ -2181,3 +2181,14 @@ SELECT view_name from timescaledb_information.continuous_aggregates WHERE hypert
 -------------------
  cagg_hypertab_ddl
 
+-- TEST continuous aggregate with functionally dependent columns
+-- name column depends on id which is in GROUP BY so name doesnt have to be
+CREATE table sensor(id int PRIMARY KEY, name text);
+CREATE TABLE sensordata(time timestamptz,id int, value float) WITH	(timescaledb.hypertable);
+NOTICE:  using column "time" as partitioning column
+CREATE MATERIALIZED VIEW cagg_sensordata WITH (tsdb.continuous) AS
+SELECT s.id, s.name,time_bucket('15 minutes', sd.time) as bucket, avg(sd.value) FROM sensordata sd JOIN sensor s USING(id) GROUP BY s.id, bucket WITH NO DATA;
+SELECT * FROM cagg_sensordata;
+ id | name | bucket | avg 
+----+------+--------+-----
+

tsl/test/expected/cagg_ddl-18.out

@@ -2181,3 +2181,14 @@ SELECT view_name from timescaledb_information.continuous_aggregates WHERE hypert
 -------------------
  cagg_hypertab_ddl
 
+-- TEST continuous aggregate with functionally dependent columns
+-- name column depends on id which is in GROUP BY so name doesnt have to be
+CREATE table sensor(id int PRIMARY KEY, name text);
+CREATE TABLE sensordata(time timestamptz,id int, value float) WITH	(timescaledb.hypertable);
+NOTICE:  using column "time" as partitioning column
+CREATE MATERIALIZED VIEW cagg_sensordata WITH (tsdb.continuous) AS
+SELECT s.id, s.name,time_bucket('15 minutes', sd.time) as bucket, avg(sd.value) FROM sensordata sd JOIN sensor s USING(id) GROUP BY s.id, bucket WITH NO DATA;
+SELECT * FROM cagg_sensordata;
+ id | name | bucket | avg 
+----+------+--------+-----
+

tsl/test/sql/cagg_ddl.sql.in

@@ -1490,3 +1490,12 @@ SELECT ddl_function();
 SELECT view_name from timescaledb_information.continuous_aggregates WHERE hypertable_name='hypertab_ddl';
 SELECT ddl_function();
 SELECT view_name from timescaledb_information.continuous_aggregates WHERE hypertable_name='hypertab_ddl';
+
+-- TEST continuous aggregate with functionally dependent columns
+-- name column depends on id which is in GROUP BY so name doesnt have to be
+CREATE table sensor(id int PRIMARY KEY, name text);
+CREATE TABLE sensordata(time timestamptz,id int, value float) WITH	(timescaledb.hypertable);
+CREATE MATERIALIZED VIEW cagg_sensordata WITH (tsdb.continuous) AS
+SELECT s.id, s.name,time_bucket('15 minutes', sd.time) as bucket, avg(sd.value) FROM sensordata sd JOIN sensor s USING(id) GROUP BY s.id, bucket WITH NO DATA;
+
+SELECT * FROM cagg_sensordata;