Horton
Menu

Re: apply_scanjoin_target_to_paths and partitionwise join

Ashutosh Bapat <ashutosh.bapat.oss@gmail.com>

View thread
From: Ashutosh Bapat <ashutosh.bapat.oss@gmail.com>
To: arne.roland@malkut.net
Cc: Jakub Wartak <jakub.wartak@enterprisedb.com>, Tomas Vondra <tomas.vondra@enterprisedb.com>, Zhihong Yu <zyu@yugabyte.com>, PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>, Tom Lane <tgl@sss.pgh.pa.us>, Julien Rouhaud <rjuju123@gmail.com>, Etsuro Fujita <etsuro.fujita@gmail.com>
Date: 2024-06-05T10:57:13Z
Lists: pgsql-hackers 
On Tue, May 28, 2024 at 7:13 AM <arne.roland@malkut.net> wrote:

> 1. I think the order by pk frac limit plans had just to similar
> performance behaviour for me to bother.
> But afaics the main point of your proposal is not related to frac plans
> at all.
>

Right.


> 2. We can't expect the optimizers to simply yield better results by
> being given more options to be wrong. (Let me give a simple example:

This patch makes our lack of cross table cross column statistics worse.
> We give it more opportunity to pick something horrible.
>

I don't see the connection between cross column statistics and this bug I
am fixing. Can  you please elaborate?


> 3. I dislike, that this patch makes much harder to debug, why no
> partitionwise join is chosen.
>
Can you please elaborate more? How does my change make debugging harder?

-- 
Best Wishes,
Ashutosh Bapat

Commits

Same data as JSON: GET /api/v1/messages/:b64id/commits the thread's linked commits as JSON, with link sources. API reference →

  1. Remove redundant SET enable_partitionwise_join = on.

  2. Don't reset the pathlist of partitioned joinrels.

  3. Allow left join removals and unique joins on partitioned tables

  4. Consider fractional paths in generate_orderedappend_paths

  5. Fix handling of targetlist SRFs when scan/join relation is known empty.