Message-ID: <4fa33caf-b5d1-6887-65a9-82ce9d92765c@enterprisedb.com>
Date: Sat, 25 Feb 2023 12:45:28 +0100
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:102.0) Gecko/20100101
 Thunderbird/102.8.0
Subject: Re: BRIN indexes vs. SK_SEARCHARRAY (and preprocessing scan keys)
Content-Language: en-US
To: Heikki Linnakangas <hlinnaka@iki.fi>,
 PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>
References: <f6aab55c-35eb-4118-a7ff-571c62e6cc39@enterprisedb.com>
 <0e1f3350-c9cf-ab62-43a5-5dae314de89c@enterprisedb.com>
 <57020b2e-d9c9-9bc7-4892-b36d9bb07563@enterprisedb.com>
 <327064f1-276e-cb45-88c2-1ff498037cf5@enterprisedb.com>
 <a312b665-79fb-4ee1-e9eb-e8ed9df5479f@iki.fi>
From: Tomas Vondra <tomas.vondra@enterprisedb.com>
In-Reply-To: <a312b665-79fb-4ee1-e9eb-e8ed9df5479f@iki.fi>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
Archived-At: <https://www.postgresql.org/message-id/4fa33caf-b5d1-6887-65a9-82ce9d92765c%40enterprisedb.com>
Precedence: bulk

On 2/24/23 22:07, Heikki Linnakangas wrote:
> I had a quick look at just the preliminary cleanup patches:
> 
>> 0001-BRIN-bloom-cleanup-20230218.patch
> 
> Looks good to me
> 
>> 0002-BRIN-minmax-multi-cleanup-20230218.patch
> 
> Looks good, although it would feel more natural to me to do it the other
> way round, and define 'matches' as 'bool matches', and use DatumGetBool.
> 

Yeah, probably. I was trying to only do the minimal change because of
(maybe) backpatching this.

> Not new with this patch, but I find the 'matches' and 'matching'
> variables a bit strange. Wouldn't it be simpler to have just one variable?
> 

True. I don't recall why we did it this way.

>> 0003-Introduce-bloom_filter_size-20230218.patch
> 
> Looks good
> 
>> 0004-Add-minmax-multi-inequality-tests-20230218.patch
> 
> Looks good
> 
>> +SELECT i/5 + mod(911 * i + 483, 25),
>> +       i/10 + mod(751 * i + 221, 41)
> 
> Peculiar formulas. Was there a particular reason for these values?
> 

No, not really. I simply wanted a random-looking data, but reproducible
and deterministic. And linear congruential generator is a simple way to
do that. I just picked a couple co-prime numbers, and that's it.


regards

-- 
Tomas Vondra
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company