Re: Should consider materializing the cheapest inner path in consider_parallel_nestloop() - Mailing list pgsql-hackers

From Tender Wang
Subject Re: Should consider materializing the cheapest inner path in consider_parallel_nestloop()
Date
Msg-id CAHewXNnSvL1M76-NKyns3F9uK1RvEBZMX8z3bQ2itpLn7ZjTfQ@mail.gmail.com
Whole thread Raw
In response to RE: Should consider materializing the cheapest inner path in consider_parallel_nestloop()  ("Fujii.Yuki@df.MitsubishiElectric.co.jp" <Fujii.Yuki@df.MitsubishiElectric.co.jp>)
Responses Re: Should consider materializing the cheapest inner path in consider_parallel_nestloop()
List pgsql-hackers


Hi. Tender.

Thank you for modification.

> From: Tender Wang <tndrwang@gmail.com>
> Sent: Tuesday, June 4, 2024 7:51 PM
>       Please add more tests.  Especially please add some negative tests;
>       current patch checks that it is safe to apply materialization. It would
>       be helpful to add tests checking that materialization is not applied
>       in both checked cases:
>       1. when inner join path is not parallel safe
>       2. when matpath is not parallel safe
>
>
>
> I added a test case that inner rel is not parallel safe. Actually,
> matpath will not create if inner rel is not parallel safe. So I didn't add test case for the second  scenario.
Is there case in which matpath is not parallel safe and inner rel is parallel safe?
If right, I think that it would be suitable to add a negative test in a such case.

I looked through create_xxx_path(), and I found that almost path.parallel_safe is assigned from RelOptiInfo.consider_parallel.
Some pathes take subpath->parallel_safe into account(e.g. Material path). In most cases, Material is parallel_safe if rel is parallel
safe. Now I haven't come up a query plan that material is un parallel-safe but rel is parallel-safe.
 

Sincerely yours,
Yuuki Fujii

--
Yuuki Fujii
Information Technology R&D Center Mitsubishi Electric Corporation




--
Tender Wang
OpenPie:  https://en.openpie.com/

pgsql-hackers by date:

Previous
From: Heikki Linnakangas
Date:
Subject: Re: Trying out read streams in pgvector (an extension)
Next
From: Bertrand Drouvot
Date:
Subject: Re: Track the amount of time waiting due to cost_delay