Re: First draft of PG 19 release notes

On Mon, Oct 26, 2026 at 11:31:06PM -0400, shihao zhong wrote:
> Hi Bruce,
>
> > Added.  Again, couldn't see it in the commit message but found it in
> > the patch.  How did you find all of these items?
>
> Thanks for taking these in.
>
> As a brief background: these items were identified while running stability
> tests
> for the PG 19 release. I built a local multi-agent LLM system to test commits
> between PG 18.6 and PG 19.
>
> It flagged 20-30 potential issues, which I then manually verified.
>
> I listed the confirmed bugs in [1], while the remaining items were non-bug
> behavior changes, which is why I shared them in that post.

Wow, I am surprised you only found these few missing items considering
how thorough your testing was. I would like to have this happen for
every major release, though I am not sure if I am capable of doing it.
Hopefully you will still be around next year. :-)

> With Lz4 change(3), I agree with your judgement. 
>
> For the NOT NULL case(4), I had worked with Tom to fix it in
> fc189eb7508.

Oh, great! Problem solved!

> > Oh, I wonder if users are parsing those?
> II actually parsed LSNs myself for a replay verifier unit test, though I'm not
> sure 
> if anyone else has a similar use case.
>
> Either way, I'm fine with leaving it out of the release notes. 

Yes, let's see what reports we get.

> > Yes, can you please write one and I will add it to the attached
> > patch and apply it.  Thanks.
>
> Patch attached. It just drops the sentence.

Applied and I sent a separate email.

> Thanks for writing the patch for me, this is my first time dealing with a
> release
> note so do not know we can also write a patch to it.
>
> Three small things:
>
> 1. io_workers was replaced by four variables, not three.(io_min_workers, 
> io_max_workers, io_worker_idle_timeout, io_worker_launch_interval)

Right, not sure how I got "three" stuck in my head.

> 2. The names in pg_wait_events are BufferShared, BufferShareExclusive
> and BufferExclusive. The upper case ones are internal names.

Yes, this one confused me. I could only find the uppercase names so I
used those. Thanks to you, I now see that the lowercase names exist in
src/backend/utils/pgstat_wait_event.c, but this is created by
src/backend/utils/activity/generate-wait_event_types.pl so is not in the
patch. I have adjusted for the lowercase versions as you suggested.

> 3. I think the porter dictionary is the 1980 Porter algorithm and was there
> before 19, so it is not the previous English stemmer. The German
> change is listed in the Snowball 3.0.0 NEWS file under "German".[2]

Huh, based on the docs we have the old Dutch version available as
dutch_porter, but no previous English or German versions available.
Doesn't this require a rebuilding of indexes that use these stemmers?

Updated patch attached.

--
Bruce Momjian <bruce(at)momjian(dot)us> https://momjian.us
EDB https://enterprisedb.com

Do not let urgent matters crowd out time for investment in the future.