Re: psql's 001_basic.pl test could fail on very slow machines

From: Oleg Tselebrovskiy <o(dot)tselebrovskiy(at)postgrespro(dot)ru>
To: Greg Sabino Mullane <htamfids(at)gmail(dot)com>
Cc: pgsql-hackers(at)postgresql(dot)org
Subject: Re: psql's 001_basic.pl test could fail on very slow machines
Date: 2026-03-10 11:01:47
Message-ID: bd9cab9439d724f935c41f6207809542@postgrespro.ru
Views: Whole Thread | Raw Message | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

Greg Sabino Mullane wrote at 2026-03-09 22:07:
> I don't think your new test is really testing the full functionality
> of \watch min, but I agree that existing test does not either. How
> about something like this?
>
> select now() as start \gset
> with x as (select now() - :'start' as howlong)
> select 123 from x where howlong < '0.5 second' \watch i=0.2 m=1

Yeah, your query fits the spirit of the test more and doesn't fail
with modified post_auth_delay. We would expect 3 rows with
this wait interval and '0.5 second' limit, but on really slow
machines we could wait more than 0.2 seconds between watches and
the test would still fail

Maybe we could use sequences since we need a query that could
self-terminate, like in attached?

Regards,
Oleg

Attachment Content-Type Size
0002_rewrite_test.patch text/x-diff 769 bytes
0002_manual_reproduction.patch text/x-diff 640 bytes

In response to

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Ranier Vilela 2026-03-10 11:09:46 Re: Simplifies checks (src/bin/pg_dump/pg_backup_archiver.c)
Previous Message Anders Åstrand 2026-03-10 10:50:42 Re: Potential security risk associated with function call