Horton

Re: Memory consumed by child SpecialJoinInfo in partitionwise join planning

Amit Langote <amitlangote09@gmail.com>

View thread View on postgr.es
From: Amit Langote <amitlangote09@gmail.com>
To: Richard Guo <guofenglinux@gmail.com>
Cc: Ashutosh Bapat <ashutosh.bapat.oss@gmail.com>, Andrei Lepikhov <a.lepikhov@postgrespro.ru>, Tomas Vondra <tomas.vondra@enterprisedb.com>, pgsql-hackers <pgsql-hackers@postgresql.org>, vignesh C <vignesh21@gmail.com>
Date: 2024-03-25T10:31:37Z
Lists: pgsql-hackers 
On Mon, Mar 25, 2024 at 7:25 PM Richard Guo <guofenglinux@gmail.com> wrote:
> On Mon, Mar 25, 2024 at 5:17 PM Amit Langote <amitlangote09@gmail.com> wrote:
>> I've pushed this now.
>>
>> When updating the commit message, I realized that you had made the
>> right call to divide the the changes around not translating the dummy
>> SpecialJoinInfos into a separate patch.  So, I pushed it like that.
>
>
> Sorry I'm late for the party.  I noticed that in commit 6190d828cd the
> comment of init_dummy_sjinfo() mentions the non-existing 'rel1' and
> 'rel2', which should be addressed.

Thanks for catching that.  Oops.

>  Also, I think we can make function
> consider_new_or_clause() call init_dummy_sjinfo() to simplify the code a
> bit.

Indeed.

> Attached is a trivial patch for the fixes.

Will apply shortly.

-- 
Thanks, Amit Langote

Commits

  1. Code review for 6190d828cd2

  2. Do not translate dummy SpecialJoinInfos for child joins

  3. Reduce memory used by partitionwise joins

  4. Remove dummy_spinlock