Feedback-ID: ia2694551:Fastmail
Date: Wed, 1 Mar 2023 19:33:01 +0100
From: Alvaro Herrera <alvherre@alvh.no-ip.org>
To: Tomas Vondra <tomas.vondra@enterprisedb.com>
Cc: Matthias van de Meent <boekewurm+postgres@gmail.com>,
	Justin Pryzby <pryzby@telsasoft.com>,
	Andres Freund <andres@anarazel.de>, Greg Stark <stark@mit.edu>,
	Zhihong Yu <zyu@yugabyte.com>,
	PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>
Subject: Re: PATCH: Using BRIN indexes for sorted output
Message-ID: <20230301183301.w3m5f2fbovm5hv5c@alvherre.pgsql>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
In-Reply-To: <5598409b-4ce8-dec2-acee-272e94c4b21b@enterprisedb.com>
Archived-At: <https://www.postgresql.org/message-id/20230301183301.w3m5f2fbovm5hv5c%40alvherre.pgsql>
Precedence: bulk

On 2023-Feb-24, Tomas Vondra wrote:

> On 2/24/23 16:14, Alvaro Herrera wrote:

> > I think a formulation of this kind has the benefit that it works after
> > BlockNumber is enlarged to 64 bits, and doesn't have to be changed ever
> > again (assuming it is correct).
> 
> Did anyone even propose doing that? I suspect this is unlikely to be the
> only place that'd might be broken by that.

True about other places also needing fixes, and no I haven't see anyone;
but while 32 TB does seem very far away to us now, it might be not
*that* far away.  So I think doing it the other way is better.

> > ... if pagesPerRange is not a whole divisor of MaxBlockNumber, I think
> > this will neglect the last range in the table.
> 
> Why would it? Let's say BlockNumber is uint8, i.e. 255 max. And there
> are 10 pages per range. That's 25 "full" ranges, and the last range
> being just 5 pages. So we get into
> 
>    prevHeapBlk = 240
>    heapBlk = 250
> 
> and we read the last 5 pages. And then we update
> 
>    prevHeapBlk = 250
>    heapBlk = (250 + 10) % 255 = 5
> 
> and we don't do that loop. Or did I get this wrong, somehow?

I stand corrected.

-- 
Álvaro Herrera               48°01'N 7°57'E  —  https://www.EnterpriseDB.com/