-- create a table to use as a basis for views and materialized views in various combinations CREATE TABLE t (id int NOT NULL PRIMARY KEY, type text NOT NULL, amt numeric NOT NULL); INSERT INTO t VALUES (1, 'x', 2), (2, 'x', 3), (3, 'y', 5), (4, 'y', 7), (5, 'z', 11); -- we want a view based on the table, too, since views present additional challenges CREATE VIEW tv AS SELECT type, sum(amt) AS totamt FROM t GROUP BY type; SELECT * FROM tv ORDER BY type; type | totamt ------+-------- x | 5 y | 12 z | 11 (3 rows) -- create a materialized view with no data, and confirm correct behavior EXPLAIN (costs off) CREATE MATERIALIZED VIEW tm AS SELECT type, sum(amt) AS totamt FROM t GROUP BY type WITH NO DATA; NOTICE: Table doesn't have 'DISTRIBUTED BY' clause -- Using column(s) named 'type' 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. QUERY PLAN ------------------------------------------------------ HashAggregate Group Key: t.type -> Redistribute Motion 3:3 (slice1; segments: 3) Hash Key: t.type -> HashAggregate Group Key: t.type -> Seq Scan on t Optimizer: Postgres query optimizer (8 rows) CREATE MATERIALIZED VIEW tm AS SELECT type, sum(amt) AS totamt FROM t GROUP BY type WITH NO DATA; SELECT relispopulated FROM pg_class WHERE oid = 'tm'::regclass; relispopulated ---------------- f (1 row) SELECT * FROM tm; ERROR: materialized view "tm" has not been populated HINT: Use the REFRESH MATERIALIZED VIEW command. REFRESH MATERIALIZED VIEW tm; SELECT relispopulated FROM pg_class WHERE oid = 'tm'::regclass; relispopulated ---------------- t (1 row) CREATE UNIQUE INDEX tm_type ON tm (type); SELECT * FROM tm; type | totamt ------+-------- y | 12 z | 11 x | 5 (3 rows) -- create various views EXPLAIN (costs off) CREATE MATERIALIZED VIEW tvm AS SELECT * FROM tv ORDER BY type; NOTICE: Table doesn't have 'DISTRIBUTED BY' clause -- Using column(s) named 'type' 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. QUERY PLAN ------------------------------------------------------------ Sort Sort Key: t.type -> HashAggregate Group Key: t.type -> Redistribute Motion 3:3 (slice1; segments: 3) Hash Key: t.type -> HashAggregate Group Key: t.type -> Seq Scan on t Optimizer: Postgres query optimizer (10 rows) CREATE MATERIALIZED VIEW tvm AS SELECT * FROM tv ORDER BY type; SELECT * FROM tvm; type | totamt ------+-------- x | 5 y | 12 z | 11 (3 rows) CREATE MATERIALIZED VIEW tmm AS SELECT sum(totamt) AS grandtot FROM tm; CREATE MATERIALIZED VIEW tvmm AS SELECT sum(totamt) AS grandtot FROM tvm; CREATE UNIQUE INDEX tvmm_expr ON tvmm ((grandtot > 0)); CREATE UNIQUE INDEX tvmm_pred ON tvmm (grandtot) WHERE grandtot < 0; CREATE VIEW tvv AS SELECT sum(totamt) AS grandtot FROM tv; EXPLAIN (costs off) CREATE MATERIALIZED VIEW tvvm AS SELECT * FROM tvv; NOTICE: Table doesn't have 'DISTRIBUTED BY' clause -- Using column(s) named 'grandtot' 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. QUERY PLAN ------------------------------------------------------------------------------------ Redistribute Motion 1:3 (slice3; segments: 1) Hash Key: (pg_catalog.sum((sum(tv.totamt)))) -> Aggregate -> Gather Motion 3:1 (slice2; segments: 3) -> Aggregate -> Subquery Scan on tv -> HashAggregate Group Key: t.type -> Redistribute Motion 3:3 (slice1; segments: 3) Hash Key: t.type -> HashAggregate Group Key: t.type -> Seq Scan on t Optimizer: Postgres query optimizer (14 rows) CREATE MATERIALIZED VIEW tvvm AS SELECT * FROM tvv; CREATE VIEW tvvmv AS SELECT * FROM tvvm; CREATE MATERIALIZED VIEW bb AS SELECT * FROM tvvmv; CREATE INDEX aa ON bb (grandtot); -- check that plans seem reasonable \d+ tvm Materialized view "public.tvm" Column | Type | Modifiers | Storage | Stats target | Description --------+---------+-----------+----------+--------------+------------- type | text | | extended | | totamt | numeric | | main | | View definition: SELECT tv.type, tv.totamt FROM tv ORDER BY tv.type; Distributed by: (type) \d+ tvm Materialized view "public.tvm" Column | Type | Modifiers | Storage | Stats target | Description --------+---------+-----------+----------+--------------+------------- type | text | | extended | | totamt | numeric | | main | | View definition: SELECT tv.type, tv.totamt FROM tv ORDER BY tv.type; Distributed by: (type) \d+ tvvm Materialized view "public.tvvm" Column | Type | Modifiers | Storage | Stats target | Description ----------+---------+-----------+---------+--------------+------------- grandtot | numeric | | main | | View definition: SELECT tvv.grandtot FROM tvv; Distributed by: (grandtot) \d+ bb Materialized view "public.bb" Column | Type | Modifiers | Storage | Stats target | Description ----------+---------+-----------+---------+--------------+------------- grandtot | numeric | | main | | Indexes: "aa" btree (grandtot) View definition: SELECT tvvmv.grandtot FROM tvvmv; Distributed by: (grandtot) -- test schema behavior CREATE SCHEMA mvschema; ALTER MATERIALIZED VIEW tvm SET SCHEMA mvschema; \d+ tvm \d+ tvmm Materialized view "public.tvmm" Column | Type | Modifiers | Storage | Stats target | Description ----------+---------+-----------+---------+--------------+------------- grandtot | numeric | | main | | Indexes: "tvmm_expr" UNIQUE, btree ((grandtot > 0::numeric)) "tvmm_pred" UNIQUE, btree (grandtot) WHERE grandtot < 0::numeric View definition: SELECT sum(tvm.totamt) AS grandtot FROM mvschema.tvm; Distributed by: (grandtot) SET search_path = mvschema, public; \d+ tvm Materialized view "mvschema.tvm" Column | Type | Modifiers | Storage | Stats target | Description --------+---------+-----------+----------+--------------+------------- type | text | | extended | | totamt | numeric | | main | | View definition: SELECT tv.type, tv.totamt FROM tv ORDER BY tv.type; Distributed by: (type) -- modify the underlying table data INSERT INTO t VALUES (6, 'z', 13); -- confirm pre- and post-refresh contents of fairly simple materialized views SELECT * FROM tm ORDER BY type; type | totamt ------+-------- x | 5 y | 12 z | 11 (3 rows) SELECT * FROM tvm ORDER BY type; type | totamt ------+-------- x | 5 y | 12 z | 11 (3 rows) REFRESH MATERIALIZED VIEW CONCURRENTLY tm; REFRESH MATERIALIZED VIEW tvm; SELECT * FROM tm ORDER BY type; type | totamt ------+-------- x | 5 y | 12 z | 24 (3 rows) SELECT * FROM tvm ORDER BY type; type | totamt ------+-------- x | 5 y | 12 z | 24 (3 rows) RESET search_path; -- confirm pre- and post-refresh contents of nested materialized views EXPLAIN (costs off) SELECT * FROM tmm; QUERY PLAN ----------------------------------- Seq Scan on tmm Optimizer: Postgres query optimizer (2 rows) EXPLAIN (costs off) SELECT * FROM tvmm; QUERY PLAN ------------------------------------------ Gather Motion 3:1 (slice1; segments: 3) -> Seq Scan on tvmm Optimizer: Postgres query optimizer (3 rows) EXPLAIN (costs off) SELECT * FROM tvvm; QUERY PLAN ------------------------------------------ Gather Motion 3:1 (slice1; segments: 3) -> Seq Scan on tvvm Optimizer: Postgres query optimizer (3 rows) SELECT * FROM tmm; grandtot ---------- 28 (1 row) SELECT * FROM tvmm; grandtot ---------- 28 (1 row) SELECT * FROM tvvm; grandtot ---------- 28 (1 row) REFRESH MATERIALIZED VIEW tmm; REFRESH MATERIALIZED VIEW CONCURRENTLY tvmm; ERROR: cannot refresh materialized view "public.tvmm" concurrently HINT: Create a unique index with no WHERE clause on one or more columns of the materialized view. REFRESH MATERIALIZED VIEW tvmm; REFRESH MATERIALIZED VIEW tvvm; EXPLAIN (costs off) SELECT * FROM tmm; QUERY PLAN ----------------------------------- Seq Scan on tmm Optimizer: Postgres query optimizer (2 rows) EXPLAIN (costs off) SELECT * FROM tvmm; QUERY PLAN ------------------------------------------ Gather Motion 3:1 (slice1; segments: 3) -> Seq Scan on tvmm Optimizer: Postgres query optimizer (3 rows) EXPLAIN (costs off) SELECT * FROM tvvm; QUERY PLAN ------------------------------------------ Gather Motion 3:1 (slice1; segments: 3) -> Seq Scan on tvvm Optimizer: Postgres query optimizer (3 rows) SELECT * FROM tmm; grandtot ---------- 41 (1 row) SELECT * FROM tvmm; grandtot ---------- 41 (1 row) SELECT * FROM tvvm; grandtot ---------- 41 (1 row) -- test diemv when the mv does not exist DROP MATERIALIZED VIEW IF EXISTS no_such_mv; NOTICE: materialized view "no_such_mv" does not exist, skipping -- make sure invalid combination of options is prohibited REFRESH MATERIALIZED VIEW CONCURRENTLY tvmm WITH NO DATA; ERROR: CONCURRENTLY and WITH NO DATA options cannot be used together -- no tuple locks on materialized views SELECT * FROM tvvm FOR SHARE; ERROR: cannot lock rows in materialized view "tvvm" -- test join of mv and view SELECT type, m.totamt AS mtot, v.totamt AS vtot FROM tm m LEFT JOIN tv v USING (type) ORDER BY type; type | mtot | vtot ------+------+------ x | 5 | 5 y | 12 | 12 z | 24 | 24 (3 rows) -- make sure that dependencies are reported properly when they block the drop DROP TABLE t; ERROR: cannot drop table t because other objects depend on it DETAIL: view tv depends on table t view tvv depends on view tv materialized view tvvm depends on view tvv view tvvmv depends on materialized view tvvm materialized view bb depends on view tvvmv materialized view mvschema.tvm depends on view tv materialized view tvmm depends on materialized view mvschema.tvm materialized view tm depends on table t materialized view tmm depends on materialized view tm HINT: Use DROP ... CASCADE to drop the dependent objects too. -- make sure dependencies are dropped and reported -- and make sure that transactional behavior is correct on rollback -- incidentally leaving some interesting materialized views for pg_dump testing BEGIN; DROP TABLE t CASCADE; NOTICE: drop cascades to 9 other objects DETAIL: drop cascades to view tv drop cascades to view tvv drop cascades to materialized view tvvm drop cascades to view tvvmv drop cascades to materialized view bb drop cascades to materialized view mvschema.tvm drop cascades to materialized view tvmm drop cascades to materialized view tm drop cascades to materialized view tmm ROLLBACK; -- some additional tests not using base tables CREATE VIEW v_test1 AS SELECT 1 moo; CREATE VIEW v_test2 AS SELECT moo, 2*moo FROM v_test1 UNION ALL SELECT moo, 3*moo FROM v_test1; \d+ v_test2 View "public.v_test2" Column | Type | Modifiers | Storage | Description ----------+---------+-----------+---------+------------- moo | integer | | plain | ?column? | integer | | plain | View definition: SELECT v_test1.moo, 2 * v_test1.moo FROM v_test1 UNION ALL SELECT v_test1.moo, 3 * v_test1.moo FROM v_test1; CREATE MATERIALIZED VIEW mv_test2 AS SELECT moo, 2*moo FROM v_test2 UNION ALL SELECT moo, 3*moo FROM v_test2; \d+ mv_test2 Materialized view "public.mv_test2" Column | Type | Modifiers | Storage | Stats target | Description ----------+---------+-----------+---------+--------------+------------- moo | integer | | plain | | ?column? | integer | | plain | | View definition: SELECT v_test2.moo, 2 * v_test2.moo FROM v_test2 UNION ALL SELECT v_test2.moo, 3 * v_test2.moo FROM v_test2; Distributed by: (moo) CREATE MATERIALIZED VIEW mv_test3 AS SELECT * FROM mv_test2 WHERE moo = 12345; SELECT relispopulated FROM pg_class WHERE oid = 'mv_test3'::regclass; relispopulated ---------------- t (1 row) DROP VIEW v_test1 CASCADE; NOTICE: drop cascades to 3 other objects DETAIL: drop cascades to view v_test2 drop cascades to materialized view mv_test2 drop cascades to materialized view mv_test3 -- test that duplicate values on unique index prevent refresh CREATE TABLE foo(a, b) AS VALUES(1, 10); CREATE MATERIALIZED VIEW mv AS SELECT * FROM foo; CREATE UNIQUE INDEX ON mv(a); INSERT INTO foo SELECT * FROM foo; REFRESH MATERIALIZED VIEW mv; ERROR: could not create unique index "mv_a_idx" DETAIL: Key (a)=(1) is duplicated. REFRESH MATERIALIZED VIEW CONCURRENTLY mv; ERROR: new data for "mv" contains duplicate rows without any null columns DETAIL: Row: (1,10) DROP TABLE foo CASCADE; NOTICE: drop cascades to materialized view mv -- make sure that all columns covered by unique indexes works CREATE TABLE foo(a, b, c) AS VALUES(1, 2, 3); CREATE MATERIALIZED VIEW mv AS SELECT * FROM foo; CREATE UNIQUE INDEX ON mv (a); CREATE UNIQUE INDEX ON mv (b); CREATE UNIQUE INDEX on mv (c); INSERT INTO foo VALUES(2, 3, 4); INSERT INTO foo VALUES(3, 4, 5); REFRESH MATERIALIZED VIEW mv; REFRESH MATERIALIZED VIEW CONCURRENTLY mv; DROP TABLE foo CASCADE; NOTICE: drop cascades to materialized view mv -- allow subquery to reference unpopulated matview if WITH NO DATA is specified CREATE MATERIALIZED VIEW mv1 AS SELECT 1 AS col1 WITH NO DATA; CREATE MATERIALIZED VIEW mv2 AS SELECT * FROM mv1 WHERE col1 = (SELECT LEAST(col1) FROM mv1) WITH NO DATA; DROP MATERIALIZED VIEW mv1 CASCADE; NOTICE: drop cascades to materialized view mv2 -- make sure that types with unusual equality tests work CREATE TABLE boxes (id serial primary key, b box); INSERT INTO boxes (b) VALUES ('(32,32),(31,31)'), ('(2.0000004,2.0000004),(1,1)'), ('(1.9999996,1.9999996),(1,1)'); CREATE MATERIALIZED VIEW boxmv AS SELECT * FROM boxes; CREATE UNIQUE INDEX boxmv_id ON boxmv (id); UPDATE boxes SET b = '(2,2),(1,1)' WHERE id = 2; REFRESH MATERIALIZED VIEW CONCURRENTLY boxmv; SELECT * FROM boxmv ORDER BY id; id | b ----+----------------------------- 1 | (32,32),(31,31) 2 | (2,2),(1,1) 3 | (1.9999996,1.9999996),(1,1) (3 rows) DROP TABLE boxes CASCADE; NOTICE: drop cascades to materialized view boxmv -- make sure that column names are handled correctly CREATE TABLE mvtest_v (i int, j int); CREATE MATERIALIZED VIEW mvtest_mv_v (ii, jj, kk) AS SELECT i, j FROM mvtest_v; -- error ERROR: too many column names were specified CREATE MATERIALIZED VIEW mvtest_mv_v (ii, jj) AS SELECT i, j FROM mvtest_v; -- ok CREATE MATERIALIZED VIEW mvtest_mv_v_2 (ii) AS SELECT i, j FROM mvtest_v; -- ok CREATE MATERIALIZED VIEW mvtest_mv_v_3 (ii, jj, kk) AS SELECT i, j FROM mvtest_v WITH NO DATA; -- error ERROR: too many column names were specified CREATE MATERIALIZED VIEW mvtest_mv_v_3 (ii, jj) AS SELECT i, j FROM mvtest_v WITH NO DATA; -- ok CREATE MATERIALIZED VIEW mvtest_mv_v_4 (ii) AS SELECT i, j FROM mvtest_v WITH NO DATA; -- ok ALTER TABLE mvtest_v RENAME COLUMN i TO x; INSERT INTO mvtest_v values (1, 2); CREATE UNIQUE INDEX mvtest_mv_v_ii ON mvtest_mv_v (ii); REFRESH MATERIALIZED VIEW mvtest_mv_v; UPDATE mvtest_v SET j = 3 WHERE x = 1; REFRESH MATERIALIZED VIEW CONCURRENTLY mvtest_mv_v; REFRESH MATERIALIZED VIEW mvtest_mv_v_2; REFRESH MATERIALIZED VIEW mvtest_mv_v_3; REFRESH MATERIALIZED VIEW mvtest_mv_v_4; SELECT * FROM mvtest_v; x | j ---+--- 1 | 3 (1 row) SELECT * FROM mvtest_mv_v; ii | jj ----+---- 1 | 3 (1 row) SELECT * FROM mvtest_mv_v_2; ii | j ----+--- 1 | 3 (1 row) SELECT * FROM mvtest_mv_v_3; ii | jj ----+---- 1 | 3 (1 row) SELECT * FROM mvtest_mv_v_4; ii | j ----+--- 1 | 3 (1 row) DROP TABLE mvtest_v CASCADE; NOTICE: drop cascades to 4 other objects DETAIL: drop cascades to materialized view mvtest_mv_v drop cascades to materialized view mvtest_mv_v_2 drop cascades to materialized view mvtest_mv_v_3 drop cascades to materialized view mvtest_mv_v_4 -- make sure that create WITH NO DATA does not plan the query (bug #13907) create materialized view mvtest_error as select 1/0 as x; -- fail ERROR: division by zero create materialized view mvtest_error as select 1/0 as x with no data; refresh materialized view mvtest_error; -- fail here ERROR: division by zero drop materialized view mvtest_error; -- make sure that matview rows can be referenced as source rows (bug #9398) CREATE TABLE v AS SELECT generate_series(1,10) AS a; CREATE MATERIALIZED VIEW mv_v AS SELECT a FROM v WHERE a <= 5; DELETE FROM v WHERE EXISTS ( SELECT * FROM mv_v WHERE mv_v.a = v.a ); SELECT * FROM v; a ---- 6 7 8 9 10 (5 rows) SELECT * FROM mv_v; a --- 1 2 3 4 5 (5 rows) DROP TABLE v CASCADE; NOTICE: drop cascades to materialized view mv_v -- make sure running as superuser works when MV owned by another role (bug #11208) CREATE ROLE user_dw; SET ROLE user_dw; CREATE TABLE foo_data AS SELECT i, md5(random()::text) FROM generate_series(1, 10) i; CREATE MATERIALIZED VIEW mv_foo AS SELECT * FROM foo_data; CREATE UNIQUE INDEX ON mv_foo (i); RESET ROLE; REFRESH MATERIALIZED VIEW mv_foo; REFRESH MATERIALIZED VIEW CONCURRENTLY mv_foo; DROP OWNED BY user_dw CASCADE; DROP ROLE user_dw; -- make sure that create WITH NO DATA works via SPI BEGIN; CREATE FUNCTION mvtest_func() RETURNS void AS $$ BEGIN CREATE MATERIALIZED VIEW mvtest1 AS SELECT 1 AS x; CREATE MATERIALIZED VIEW mvtest2 AS SELECT 1 AS x WITH NO DATA; END; $$ LANGUAGE plpgsql; SELECT mvtest_func(); mvtest_func ------------- (1 row) SELECT * FROM mvtest1; x --- 1 (1 row) SELECT * FROM mvtest2; ERROR: materialized view "mvtest2" has not been populated HINT: Use the REFRESH MATERIALIZED VIEW command. ROLLBACK;