Re: [PATCH] Space reservation v02

From: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
To: Gregory Stark <stark(at)enterprisedb(dot)com>
Cc: Heikki Linnakangas <heikki(dot)linnakangas(at)enterprisedb(dot)com>, Zdenek Kotala <Zdenek(dot)Kotala(at)Sun(dot)COM>, Bruce Momjian <bruce(at)momjian(dot)us>, "pgsql-hackers\(at)postgresql(dot)org" <pgsql-hackers(at)postgresql(dot)org>
Subject: Re: [PATCH] Space reservation v02
Date: 2009-01-30 17:14:42
Message-ID: 25192.1233335682@sss.pgh.pa.us
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

Gregory Stark <stark(at)enterprisedb(dot)com> writes:
> Well having a column in pg_class does have some advantages. Like, you could
> look at the value from an sql session more easily. And if there are operations
> which we know are unsafe -- such as adding columns -- we could clear it from
> the server side easily.

Why would there be any unsafe operations? Surely the patch would add
sufficient logic to prevent the old version from de-fixing any page
that had already been fixed. If this is not so, the entire concept
is broken, because you're still going to have to go to single-user mode
for a long time to make sure that the whole database is in good shape.

On the whole I agree with Heikki's earlier criticism: this is all
about guessing the future, and the odds seem high that the actual
requirements will not be what you designed for anyway.

regards, tom lane

In response to

Browse pgsql-hackers by date

  From Date Subject
Next Message Sam Mason 2009-01-30 17:15:28 Re: using composite types in insert/update
Previous Message Bruce Momjian 2009-01-30 17:06:53 Re: [PATCH] Space reservation v02