Re: Optimize SnapBuildPurgeOlderTxn: use in-place compaction instead of temporary array

From: Xuneng Zhou <xunengzhou(at)gmail(dot)com>
To: Neil Chen <carpenter(dot)nail(dot)cz(at)gmail(dot)com>
Cc: pgsql-hackers <pgsql-hackers(at)lists(dot)postgresql(dot)org>, Masahiko Sawada <sawada(dot)mshk(at)gmail(dot)com>, Michael Paquier <michael(at)paquier(dot)xyz>, Kirill Reshke <reshkekirill(at)gmail(dot)com>
Subject: Re: Optimize SnapBuildPurgeOlderTxn: use in-place compaction instead of temporary array
Date: 2026-01-08 15:54:35
Message-ID: CABPTF7UKaXP+1-jMg7yXWrj+C07ZvDitDpLF6UeCQUAFGO4nYw@mail.gmail.com
Views: Whole Thread | Raw Message | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

Hi Neil,

On Thu, Jan 8, 2026 at 4:49 PM Neil Chen <carpenter(dot)nail(dot)cz(at)gmail(dot)com> wrote:
>
> Hi Xuneng,
>
> On Thu, Jan 8, 2026 at 4:15 PM Xuneng Zhou <xunengzhou(at)gmail(dot)com> wrote:
>>
>>
>> v3-0001 fixes a critical issue where the snapshot->xip array in
>> SnapBuildBuildSnapshot might not be sorted before reaching the
>> consistent state. Sorry for the noise here.
>>
>
> I’ve given this patch a cursory review, and it looks good overall.
> Considering the impact of this change, should we add a regression test for it, or provide a unit test to ensure the correctness of the modification?
> I think it would be more receptive to merging such a "subtle performance optimization" only when its correctness is fully guaranteed.

Thanks for your review. I think we can add a tap test for it.

--
Best,
Xuneng

In response to

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Fujii Masao 2026-01-08 16:02:31 Re: [BUG] CRASH: ECPGprepared_statement() and ECPGdeallocate_all() when connection is NULL
Previous Message Tom Lane 2026-01-08 15:42:09 Re: Proposal: SELECT * EXCLUDE (...) command