From: Tom Lane <tgl@sss.pgh.pa.us>
To: Melanie Plageman <melanieplageman@gmail.com>
cc: Tomas Vondra <tomas.vondra@enterprisedb.com>,
        Thomas Munro <thomas.munro@gmail.com>,
        Heikki Linnakangas <hlinnaka@iki.fi>,
        Andres Freund <andres@anarazel.de>,
        Pg Hackers <pgsql-hackers@postgresql.org>,
        Nazir Bilal Yavuz <byavuz81@gmail.com>
Subject: Re: BitmapHeapScan streaming read user and prelim refactoring
In-reply-to: <20240406134011.hjo3e2336sgcghfp@liskov>
References: <20240327193750.3mlcmzqondpj27xe@liskov>
 <7e98781c-34e0-4c02-8e5b-0500aa074271@enterprisedb.com>
 <CA+hUKGJtm_gkmW_h_02-Q9ZRcG3yOx2uzVqbCTfz7YPnTfs+DA@mail.gmail.com>
 <8330278b-9615-4030-8626-51ce714a7f54@enterprisedb.com>
 <20240331154551.lty4mundyoyhtixw@liskov>
 <20240403225759.sq5hmxdmbalroosy@liskov>
 <95d9471f-ead4-4215-b739-f6d6f4c19497@enterprisedb.com>
 <20240405080634.d7c23f46xjhhxf5q@liskov>
 <20240405235330.v4didvweb6isodtk@liskov>
 <a76f657a-5ce9-4467-b832-356c881db981@enterprisedb.com>
 <20240406134011.hjo3e2336sgcghfp@liskov>
Comments: In-reply-to Melanie Plageman <melanieplageman@gmail.com>
	message dated "Sat, 06 Apr 2024 09:40:11 -0400"
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <1939304.1712415547.1@sss.pgh.pa.us>
Content-Transfer-Encoding: quoted-printable
Date: Sat, 06 Apr 2024 10:59:07 -0400
Message-ID: <1939305.1712415547@sss.pgh.pa.us>
Archived-At: 
 <https://www.postgresql.org/message-id/1939305.1712415547%40sss.pgh.pa.us>
Precedence: bulk

Melanie Plageman <melanieplageman@gmail.com> writes:
> On Sat, Apr 06, 2024 at 02:51:45AM +0200, Tomas Vondra wrote:
>> * The one question I'm somewhat unsure about is why Tom chose to use th=
e
>> "wrong" recheck flag in the 2017 commit, when the correct recheck flag
>> is readily available. Surely that had a reason, right? But I can't thin=
k
>> of one ...

> See above.

Hi, I hadn't been paying attention to this thread, but Melanie pinged
me off-list about this question.  I think it's just a flat-out
oversight in 7c70996eb.  Looking at the mailing list thread
(particularly [1][2]), it seems that Alexander hadn't really addressed
the question of when to prefetch at all, but just skipped prefetch if
the current page was skippable:

+		/*
+		 * If we did not need to fetch the current page,
+		 * we probably will not need to fetch the next.
+		 */
+		return;

It looks like I noticed that we could check the appropriate VM bits,
but failed to notice that we could easily check the appropriate
recheck flag as well.  Feel free to change it.

			regards, tom lane

[1] https://www.postgresql.org/message-id/a6434d5c-ed8d-b09c-a7c3-b2d1677e=
35b3%40postgrespro.ru
[2] https://www.postgresql.org/message-id/5974.1509573988%40sss.pgh.pa.us