Re: alter table tablename add column - breaks pl/pgsql function returns tablename - Mailing list pgsql-hackers

From Palle Girgensohn
Subject Re: alter table tablename add column - breaks pl/pgsql function returns tablename
Date
Msg-id 4617A076-4084-4CB0-96E7-C09FA9795EDE@pingpong.net
Whole thread Raw
In response to Re: alter table tablename add column - breaks pl/pgsql function returns tablename  (Tom Lane <tgl@sss.pgh.pa.us>)
Responses Re: alter table tablename add column - breaks pl/pgsql function returns tablename  (Tom Lane <tgl@sss.pgh.pa.us>)
List pgsql-hackers

5 nov 2012 kl. 22:23 skrev Tom Lane <tgl@sss.pgh.pa.us>:

> Palle Girgensohn <girgen@pingpong.net> writes:
>> Please note that this problem does not go away by disconnecting and reconnecting, and other sessions get the error
immediately,so the claim that it is bound to a session is false. 
>
> Huh?  The test case you provided certainly doesn't exhibit any such
> behavior.  I get
>
> regression=# SELECT * FROM test_func();
> ERROR:  wrong record type supplied in RETURN NEXT
> CONTEXT:  PL/pgSQL function test_func() line 6 at RETURN NEXT
> regression=# \c -
> You are now connected to database "regression" as user "postgres".
> regression=# SELECT * FROM test_func();
> id | foo
> ----+-----
>  1 |
> (1 row)
>
>            regards, tom lane


Ah, sorry. Other sessions get the error immediately as well though. Would input parameters matter, or is it just the
returntype? I'll see if I can find a test case that breaks permanently, but I'm probably mistaken about that bit then.  


pgsql-hackers by date:

Previous
From: Josh Berkus
Date:
Subject: Re: What are the advantages of not being able to access multiple databases with one connection?
Next
From: Jeff Janes
Date:
Subject: Re: Pg_upgrade speed for many tables