EphemeralNamedRelation and materialized view - Mailing list pgsql-hackers

From Yugo Nagata
Subject EphemeralNamedRelation and materialized view
Date
Msg-id 20240726160714.e74d0db579f2c017e1ca0b7e@sraoss.co.jp
Whole thread Raw
Responses Re: EphemeralNamedRelation and materialized view
Re: EphemeralNamedRelation and materialized view
List pgsql-hackers
Hi,

While looking into the commit b4da732fd64e936970f38c792f8b32c4bdf2bcd5,
I noticed that we can create a materialized view using Ephemeral Named
Relation in PostgreSQL 16 or earler. 


postgres=# create table tbl (i int);
CREATE TABLE
                                                     ^
postgres=# create or replace function f() returns trigger as $$ begin 
 create materialized view mv as select * from enr; return new; end; $$ language plpgsql;
CREATE FUNCTION

postgres=# create trigger trig after insert on tbl referencing new table as enr execute function f();
CREATE TRIGGER

postgres=# insert into tbl values (10);

postgres=# \d
             List of relations
 Schema | Name |       Type        | Owner  
--------+------+-------------------+--------
 public | mv   | materialized view | yugo-n
 public | tbl  | table             | yugo-n
(2 rows)


We cannot refresh or get the deinition of it, though.

postgres=# refresh materialized view mv;
ERROR:  executor could not find named tuplestore "enr"

postgres=# \d+ mv
ERROR:  unrecognized RTE kind: 7

In PostgreSQL 17, materialized view using ENR cannot be created 
because queryEnv is not pass to RefreshMatViewByOid introduced by b4da732fd64.
When we try to create it, the  error is raised.

 ERROR: executor could not find named tuplestore "enr"

Although it is hard to imagine users actually try to create materialized view
using ENR, how about prohibiting it even in PG16 or earlier by passing NULL
as queryEnv arg in CreateQueryDesc to avoid to create useless matviews accidentally,
as the attached patch?


Regards,
Yugo Nagata

-- 
Yugo Nagata <nagata@sraoss.co.jp>

Attachment

pgsql-hackers by date:

Previous
From: Junwang Zhao
Date:
Subject: Re: [BUG] Fix DETACH with FK pointing to a partitioned table fails
Next
From: Tatsuo Ishii
Date:
Subject: Re: warning: dereferencing type-punned pointer