Re: BUG #18964: `ALTER DATABASE ... RESET ...` fails to reset extension parameters that no longer exist

From: Laurenz Albe <laurenz(dot)albe(at)cybertec(dot)at>
To: Nathan Bossart <nathandbossart(at)gmail(dot)com>
Cc: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>, mert(at)futo(dot)org, pgsql-bugs(at)lists(dot)postgresql(dot)org
Subject: Re: BUG #18964: `ALTER DATABASE ... RESET ...` fails to reset extension parameters that no longer exist
Date: 2025-07-24 21:06:57
Message-ID: 8902b6f8d5c19f09f4a052295b9de41d9f6970b5.camel@cybertec.at
Views: Whole Thread | Raw Message | Download mbox | Resend email
Thread:
Lists: pgsql-bugs

On Thu, 2025-07-24 at 11:30 -0500, Nathan Bossart wrote:
> On Tue, Jul 22, 2025 at 04:27:24AM +0200, Laurenz Albe wrote:
>
> > There is still one piece missing in my opinion:
> >
> > ALTER SYSTEM RESET testext.swap_limit;
> > ERROR: invalid configuration parameter name "testext.swap_limit"
> > DETAIL: "testext" is a reserved prefix.
> >
> > I think that this case should work like the others.
>
> Good catch. This should be fixed in v3.

Check.

> > I'd like to see regression tests for this, but I am not sure how
> > to best devise them.
> > One idea would be to stick them into the regression tests of some
> > contrib module, even though it is not the perfect place.
>
> Added in v3.

Great! The regression test works fine on my Linux machine.

> --- /dev/null
> +++ b/contrib/auto_explain/sql/alter_reset.sql
> [...]
> +SELECT current_database() AS datname \gset
> +CREATE ROLE regress_ae_role;
> +
> +ALTER DATABASE :datname SET auto_explain.bogus = 1;
> +ALTER ROLE regress_ae_role SET auto_explain.bogus = 1;
> +ALTER ROLE regress_ae_role IN DATABASE :datname SET auto_explain.bogus = 1;
> +ALTER SYSTEM SET auto_explain.bogus = 1;
> +
> +LOAD 'auto_explain';
> +
> +ALTER DATABASE :datname RESET auto_explain.bogus;
> +ALTER ROLE regress_ae_role RESET auto_explain.bogus;
> +ALTER ROLE regress_ae_role IN DATABASE :datname RESET auto_explain.bogus;

That is perhaps a rediculous quibble, but shouldn't that be :"datname"?
I guess the regression test database will always have a proper name...

Anyway, I'll mark the patch as "ready for committer".

Yours,
Laurenz Albe

In response to

Responses

Browse pgsql-bugs by date

  From Date Subject
Next Message Tom Lane 2025-07-24 21:18:50 Re: BUG #18998: No materialized views in INFORMATION_SCHEMA.TABLE_PRIVILEGES
Previous Message Fujii Masao 2025-07-24 18:50:07 Re: Unexpected Standby Shutdown on sync_replication_slots change