Feedback-ID: i675e48f3:Fastmail
MIME-Version: 1.0
Date: Wed, 25 Mar 2026 13:58:06 -0400
From: "Greg Burd" <greg@burd.me>
To: "Nathan Bossart" <nathandbossart@gmail.com>
Cc: "Jeff Davis" <pgsql@j-davis.com>,
 pgsql-hackers <pgsql-hackers@postgresql.org>
Message-Id: <440e2d60-60e2-4e7d-9ca2-9175599c5a07@app.fastmail.com>
In-Reply-To: <acQYVfzHoL4lVTUE@nathan>
References: <7b8681a578b5fe77103948eeb7b5cdd80fabad5d.camel@j-davis.com>
 <4f48f75d-4f2a-4240-b66d-597517796e02@app.fastmail.com>
 <3fba3f5671eddb221ba38f5a12acbe7cad27edf3.camel@j-davis.com>
 <427bfbda-c901-49e4-b725-8ddc41bec23d@app.fastmail.com>
 <887b1ee62794553a95c3578fcdc0cc1831e68b27.camel@j-davis.com>
 <d7489d3e-2cbd-4b0f-b662-b5c4386a3f1e@app.fastmail.com>
 <acGI-wnW4NxS87e0@nathan>
 <db765e0d-4674-4547-9a9d-6d4d9a0a123c@app.fastmail.com>
 <acLppkl4hutHhiuH@nathan>
 <4040d3a4-01c2-4f24-9025-57cfc19aea57@app.fastmail.com>
 <acQYVfzHoL4lVTUE@nathan>
Subject: Re: Expanding HOT updates for expression and partial indexes
Content-Type: text/plain
Content-Transfer-Encoding: 7bit
Archived-At: 
 <https://www.postgresql.org/message-id/440e2d60-60e2-4e7d-9ca2-9175599c5a07%40app.fastmail.com>
Precedence: bulk


On Wed, Mar 25, 2026, at 1:16 PM, Nathan Bossart wrote:
> I just spoke to Greg off-list and wanted to share my current thoughts on
> the list as well.  In short, while we feel that the patch is in decent
> shape and seems to be performance neutral (or maybe even positive in some
> cases), it obviously doesn't accomplish $subject, and only a couple of
> folks have looked at it in depth.  Furthermore, if this patch was committed
> and someone did find a problem, it'd be hard to justify anything except a
> revert.  So, it's probably better to keep working on the full patch set and
> try to get $subject committed much earlier in the development cycle.

Thanks Nathan,

I 100% agree with that.

As I think about this thread and the current patch vs the goals in $subject, it's just not there yet.  There are no changes in the v38 set that enable HOT updates in new cases.  What is in the v38 patch set is solid and does what it claims to do, has no performance regressions (I can find) and in cases where UPDATEs target many rows shows a performance improvement, but is that enough?  More so, is now the right time?  It's the end of the development cycle and we're about to freeze so I think the answer is obviously no.

I do have more of the work nearly ready to layer back on top of this the changes to HOT for expression indexes (and more).  And I'm confident that given the chance PHOT could layer on top of this as well.  But those are not done yet. :)

> If someone thinks that we should seriously consider committing this for
> v19, please let us know.

/me eats popcorn and waits for anyone interested to chime in...

> -- 
> nathan

If no one chimes in then I'll get this in shape for the first commit fest of the next cycle. :)

best.

-greg