MIME-Version: 1.0
References: <E1tsO99-00277d-1W@gemulon.postgresql.org>
 <GV0P278MB041968707F13439C8E670752D2D02@GV0P278MB0419.CHEP278.PROD.OUTLOOK.COM>
In-Reply-To: 
 <GV0P278MB041968707F13439C8E670752D2D02@GV0P278MB0419.CHEP278.PROD.OUTLOOK.COM>
From: Melanie Plageman <melanieplageman@gmail.com>
Date: Wed, 12 Mar 2025 12:07:24 -0400
Message-ID: 
 <CAAKRu_ZHECM5htitn09XyssxmQF9Jkpw+7EK4g5_VqEpkfQPBg@mail.gmail.com>
Subject: Re: pgsql: Modularize log_connections output
To: "Daniel Westermann (DWE)" <daniel.westermann@dbi-services.com>
Cc: "pgsql-committers@lists.postgresql.org"
 <pgsql-committers@lists.postgresql.org>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Archived-At: 
 <https://www.postgresql.org/message-id/CAAKRu_ZHECM5htitn09XyssxmQF9Jkpw%2B7EK4g5_VqEpkfQPBg%40mail.gmail.com>
Precedence: bulk

Thanks for taking a look!

On Wed, Mar 12, 2025 at 11:50=E2=80=AFAM Daniel Westermann (DWE)
<daniel.westermann@dbi-services.com> wrote:
>
>
> great, thanks for this. One question:
>
> >The current log_connections options are 'receipt', 'authentication', and
> >'authorization'. The empty string disables all connection logging. 'all'
> >enables all available connection logging.
>
> ... shouldn't this be added to tab completion as well?
>
> postgres=3D# alter system set log_connections TO DEFAULT

Because it is now a string GUC, there isn't a facility for tab
completion. Enum and boolean GUCs have a defined set of options which
can be provided as options for tab completion. Because string GUCs
don't have to have a set of defined options, they don't have tab
completion (AFAIK). For example, you couldn't have a defined set of
options for log_line_prefix.

There are a subset of the string GUCs that do have predefined options
which they validate in their check hooks. Many of these are GUCs which
can accept some combination of options -- so they don't fit well into
the enum GUC type (for example, debug_io_direct).

While writing this feature, I did consider if we should add a new GUC
type PGC_SET [1] which would back GUCs that accept any subset of a
predefined set of options. This would allow us to have tab completion.
I didn't end up doing this because I felt tab completion was the only
advantage it would provide this patch and it would be a big code
change, but it is definitely something to consider in the future.

- Melanie

[1] https://www.postgresql.org/message-id/CAAKRu_a5-7sUP%2BQ6YD5emQYS1w7ffB=
DUNf-NMbcxR-dpOdGehw%40mail.gmail.com