From: "Stupor Genius" <stuporg@erols.com>
To: "Hackers" <pgsql-hackers@postgresql.org>
Subject: RE: [HACKERS] Lots 'o patches
Date: Mon, 1 Jun 1998 20:44:23 -0400
Message-ID: <001001bd8dbf$964de840$4299accf@darren>
MIME-Version: 1.0
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
In-Reply-To: <199806011424.KAA08421@candle.pha.pa.us>
Importance: Normal

>
> That is going to be difficult to do.  We used to have some SQL scripts
> that could make the required database changes, but when system table
> structure changes, I can't imagine how we would migrate that without a
> dump/reload.  I suppose we could keep the data/index files with user data,
> run initdb, and move the data files back, but we need the system table
> info reloaded into the new system tables.

If the tuple header info doesn't change, this doesn't seem that tough.
Just do a dump the pg_* tables and reload them.  The system tables are
"small" compared to the size of user data/indexes, no?

Or is there some extremely obvious reason that this is harder than it
seems?

But then again, what are the odds that changes for a release will only
affect system tables so not to require a data dump?  Not good I'd say.

darrenk