Hi,

On Mon, Nov 16, 2026 at 10:56:09AM -0600, Sami Imseih wrote:
> Attached is v20 which addresses the points raised by
> both Zsolt and Bertrand above.

Thanks!

> 2/ n_ins_since_vacuum after TRUNCATE plus an aborted savepoint is
> fixed. I added a test for this.

It looks like that it now makes the reverse order produce the wrong result, for
example:

"
BEGIN;
SAVEPOINT s;
INSERT INTO t SELECT generate_series(1, 5);
ROLLBACK TO SAVEPOINT s;

SELECT pg_stat_force_next_flush();

TRUNCATE t;
COMMIT;

SELECT pg_stat_force_next_flush();
"

produces 5 for n_ins_since_vacuum, while 0 is expected.

Regards,

-- 
Bertrand Drouvot
PostgreSQL Contributors Team
RDS Open Source Databases
Amazon Web Services: https://aws.amazon.com


Reply via email to