From: | Zdenek Kotala <Zdenek(dot)Kotala(at)Sun(dot)COM> |
---|---|
To: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
Cc: | Heikki Linnakangas <heikki(dot)linnakangas(at)enterprisedb(dot)com>, PostgreSQL-development <pgsql-hackers(at)postgresql(dot)org> |
Subject: | Re: pg_upgrade: How to deal with toast |
Date: | 2008-11-20 18:21:23 |
Message-ID: | 4925AAA3.80302@sun.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
Tom Lane napsal(a):
> Zdenek Kotala <Zdenek(dot)Kotala(at)Sun(dot)COM> writes:
>> Heikki Linnakangas napsal(a):
>>> Perhaps we should just add the new attid attribute to the toast table,
>>> but mark it as nullable?
>
>> Hmm, It seems to me as a good idea.
>
> No, it's a really horrid idea. Nullable attributes complicate the C
> code, and what in the world are we buying with it anyway? Just decide
> what the field should contain and put it in there.
The problem what we try to solve is to perform this change during upgrade from
8.3->8.4. Extra value is a problem because it requires extra space and there is
not free space. It is temporal solution(hack) for 8.3->8.4.
Another thing what we can do is to perform "fake" page conversion of heap which
will retoast a toasted value which are present on heap tuples. The toasted table
will contains two kind of tuples, but in normal situation only converted tuples
should be accessed.
Zdenek
From | Date | Subject | |
---|---|---|---|
Next Message | Bruce Momjian | 2008-11-20 18:21:59 | Re: TODO list request: FK to unique expression indexes |
Previous Message | Tom Lane | 2008-11-20 17:28:01 | Re: Re: [COMMITTERS] pgsql: Silence compiler warning about ignored return value. |