Subject: Re: PostgreSQL question
From: Guillaume Lelarge <guillaume@lelarge.info>
To: Mridul Mathew <mridulmathew@gmail.com>
Cc: pgsql-admin@postgresql.org
Date: Mon, 26 Sep 2011 20:44:50 +0200
In-Reply-To: 
 <CAFm5QJxD5tThQKXayMRJ1YQ5UR3MaHAcJ4dHE1kd7tYCkj4f6g@mail.gmail.com>
References: 
 <CAFm5QJxD5tThQKXayMRJ1YQ5UR3MaHAcJ4dHE1kd7tYCkj4f6g@mail.gmail.com>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: 7bit
Message-ID: <1317062690.2085.1.camel@localhost.localdomain>
Mime-Version: 1.0

On Mon, 2011-09-26 at 16:06 +0530, Mridul Mathew wrote:
> Hello,
> 
> I wanted to take a moment here and reach out to the community with a
> question. How does postgreSQL handle schema switching in the database? I
> have a test scenario.
> 
> ##################
> 
> vpn2=> show search_path;
>  search_path
> -------------
>  mridul
> (1 row)
> 
> CREATE OR REPLACE PACKAGE mridul.TEST_PKG
> IS
> PROCEDURE execute_for_all_schema();
> end TEST_PKG;
> 
> CREATE OR REPLACE PACKAGE BODY mridul.TEST_PKG
> IS
> 
>     PROCEDURE execute_for_all_schema() IS
> 
>         CURSOR c_schema_names IS
>         SELECT name
>         FROM mridul.schema_config
>         WHERE status = 'ACTIVE';
> 
>     BEGIN
> 
>         FOR dat_schema_names IN c_schema_names LOOP
> -----    EXECUTE IMMEDIATE 'ALTER SESSION SET CURRENT_SCHEMA = '||
> dat_schema_names.name;
>         EXECUTE IMMEDIATE 'SET SEARCH_PATH = '|| dat_schema_names.name;
>         EXECUTE IMMEDIATE 'insert into A values (1)';
>         commit;
>         END LOOP;
>         EXCEPTION
>         WHEN OTHERS THEN
>             RAISE;
> 
>     END;
> end    TEST_PKG;
> 
> vpn2=> select * from mridul.schema_config;
>  id | name  | description | status | last_device_sync_date
> ----+-------+-------------+--------+-----------------------
>   1 | TEST  | TESTING     | ACTIVE | 22-SEP-11 18:28:11
>   6 | TEST1 | TESTING5    | ACTIVE | 22-SEP-11 11:46:09
>   2 | TEST2 | TESTING2    | ACTIVE | 23-SEP-11 11:54:35
> (3 rows)
> 
> vpn2=> show search_path;
>  search_path
> -------------
>  mridul
> (1 row)
> 
> vpn2=> select count(*) from test.a;
>  count
> -------
>      0
> (1 row)
> 
> 
> vpn2=> select count(*) from test1.a;
>  count
> -------
>      0
> (1 row)
> 
> 
> vpn2=> select count(*) from test2.a;
>  count
> -------
>      0
> (1 row)
> 
> vpn2=> exec mridul.TEST_PKG.execute_for_all_schema;
> 
> EDB-SPL Procedure successfully completed
> vpn2=> select count(*) from test.a;
>  count
> -------
>      1
> (1 row)
> 
> 
> vpn2=> select count(*) from test1.a;
>  count
> -------
>      1
> (1 row)
> 
> 
> vpn2=> select count(*) from test2.a;
>  count
> -------
>      1
> (1 row)
> 
> If I convert the 'insert into A ...' part from a dynamic dml to a normal
> insert, and then execute the procedure, there are 3 rows inserted into
> test.a and none go into test1.a nor test2.a. However, if I echo the
> search_path after executing the proc, it displays the last schema's name
> (meaning, it has switched the schemas, but the insert has gone into only the
> first schema).
> 

This isn't a PostgreSQL question, this is an EnterpriseDB one. Packages
are not part of the community PostgreSQL distribution. "EXECUTE
IMMEDIATE" is also not part of the community PostgreSQL distribution.

So I guess you need to ask them.


-- 
Guillaume
  http://blog.guillaume.lelarge.info
  http://www.dalibo.com