BUG #8698: cast and view - Mailing list pgsql-bugs

From balazs@obiserver.hu
Subject BUG #8698: cast and view
Date
Msg-id E1VvATZ-0004mC-B4@wrigleys.postgresql.org
Whole thread Raw
Responses Re: BUG #8698: cast and view
List pgsql-bugs
The following bug has been logged on the website:

Bug reference:      8698
Logged by:          Balazs Peter Odor
Email address:      balazs@obiserver.hu
PostgreSQL version: 9.3.2
Operating system:   Debian Linux
Description:

CREATE SCHEMA test;
CREATE TYPE test.customcompositetype AS (
  x text,
  y text
);
CREATE TABLE test.customtable (
  compositetype test.customcompositetype
);
INSERT INTO test.customtable (compositetype)
  VALUES ('(x1,y1)'), ('(x2,y2)');


SELECT * FROM test.customtable;
 compositetype
---------------
 (x1,y1)
 (x2,y2)
(2 rows)


CREATE VIEW test.customview as
  SELECT compositetype, compositetype::text AS compositetype_text
  FROM test.customtable;


SELECT * FROM test.customview;
 compositetype | compositetype_text
---------------+--------------------
 (x1,y1)       | (x1,y1)
 (x2,y2)       | (x2,y2)
(2 rows)


Ok, lets create a cast to customcompositetype to text:


CREATE FUNCTION test.customcompositetype_text(test.customcompositetype)
  RETURNS text AS
$$SELECT ($1).x || ' ' || ($1).y;$$
  IMMUTABLE
  LANGUAGE SQL
;
CREATE CAST (test.customcompositetype AS text)
  WITH FUNCTION test.customcompositetype_text(test.customcompositetype);


SELECT * FROM test.customview;
 compositetype | compositetype_text
---------------+--------------------
 (x1,y1)       | (x1,y1)
 (x2,y2)       | (x2,y2)
(2 rows)


What?? :(


Refresh the view:


CREATE OR REPLACE VIEW test.customview as
  SELECT compositetype, compositetype::text AS compositetype_text
  FROM test.customtable;


SELECT * FROM test.customview;
 compositetype | compositetype_text
---------------+--------------------
 (x1,y1)       | x1 y1
 (x2,y2)       | x2 y2
(2 rows)


Whats wrong? Why need recreate the view??

pgsql-bugs by date:

Previous
From: "Sylvester, Peter A."
Date:
Subject: Re: BUG #8687: HashAggregate runs out of memory
Next
From: Alvaro Herrera
Date:
Subject: Re: BUG #8470: 9.3 locking/subtransaction performance regression