Re: Dead code with short varlenas in toast_save_datum()

Michael Paquier <michael@paquier.xyz>

From: Michael Paquier <michael@paquier.xyz>

To: Nikhil Kumar Veldanda <veldanda.nikhilkumar17@gmail.com>

Cc: Postgres hackers <pgsql-hackers@lists.postgresql.org>

Date: 2025-08-17T06:26:20Z

Lists: pgsql-hackers

Commits

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

Add isolation test for TOAST value reuse during CLUSTER
- f57e214d1cbb 19 (unreleased) landed
Add SQL test for TOAST value allocations on rewrite
- 69f75d671475 19 (unreleased) landed
Add regression test for short varlenas saved in TOAST relations
- 225ebfe30a1a 19 (unreleased) landed

On Sat, Aug 16, 2025 at 02:34:02AM -0700, Nikhil Kumar Veldanda wrote:
> Thanks for the guidance. I’ve updated the isolation test to use a CTAS
> capturing (id, chunk_id) pre-CLUSTER and added a post-CLUSTER join to
> verify the chunk IDs match. I also reduced the number of tuples to 1.
> 
> Please find the attached patch for your review. Thanks

Cool, thanks for the new patch.  The structure looks OK.  There were
two things that itched me a bit:
- Instead of reporting the number of tuples where the chunk IDs do not
match across the rewrites, I have rewritten the query to report the
serial IDs, instead.
- Making sure that the CTAS holds some data, matching with the number
of tuples inserted.  A count(*) was enough for that. 
--
Michael