From: Tom Lane <tgl@sss.pgh.pa.us>
To: Bruce Momjian <bruce@momjian.us>
cc: Thomas Munro <thomas.munro@gmail.com>,
        Bertrand Drouvot <bertranddrouvot.pg@gmail.com>,
        Nathan Bossart <nathandbossart@gmail.com>, adam@labkey.com,
        pgsql-bugs@lists.postgresql.org
Subject: Re: BUG #18711: Attempting a connection with a database name longer
 than 63 characters now fails
In-reply-to: <Z05Ey-8ABqWUQGYX@momjian.us>
References: <Z0Dmw1ryKH8bEkyo@nathan> <Z0c9XVU6aTd8neRr@nathan>
 <1865227.1732723416@sss.pgh.pa.us>
 <CA+hUKGKKNAc599Vp7kFAnLE1=V=ceYujz_YQoSNrvNFGaJ6i7w@mail.gmail.com>
 <2094158.1732812306@sss.pgh.pa.us>
 <CA+hUKGKDC7tKMZ1v0JGH5D23F-=ADf-3UfcriVepqoi7Q_SKgQ@mail.gmail.com>
 <Z0mUMnVakLOizwID@ip-10-97-1-34.eu-west-3.compute.internal>
 <CA+hUKG+uJGt2tks0jyS=Gkjd0Uig-9cORVm-_igp50m7QwKrWw@mail.gmail.com>
 <Z048JXV9f8GczAUa@momjian.us> <1692919.1733181379@sss.pgh.pa.us>
 <Z05Ey-8ABqWUQGYX@momjian.us>
Comments: In-reply-to Bruce Momjian <bruce@momjian.us>
	message dated "Mon, 02 Dec 2024 18:37:47 -0500"
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <1710258.1733190061.1@sss.pgh.pa.us>
Date: Mon, 02 Dec 2024 20:41:01 -0500
Message-ID: <1710259.1733190061@sss.pgh.pa.us>
Archived-At: 
 <https://www.postgresql.org/message-id/1710259.1733190061%40sss.pgh.pa.us>
Precedence: bulk

Bruce Momjian <bruce@momjian.us> writes:
> On Mon, Dec  2, 2024 at 06:16:19PM -0500, Tom Lane wrote:
>> Reality is that using multiple encodings in shared catalogs just plain
>> doesn't work very well.  Even something as simple as "select datname
>> from pg_database" is likely to give garbage if there are entries in
>> encodings that don't match your current database's encoding.  What
>> Thomas is proposing is to formalize and then enforce the two usage
>> patterns that we know work okay.

> I have not heard of anyone complaining about the problem of viewing the
> shared catalog  columns so I didn't consider that benefit.

That's probably because they've learned the hard way to use one of
the two usage patterns that actually work.

Also, I think we have some code around pg_stat_activity that
suppresses possibly-wrongly-encoded data from other DBs (in query
text, not only user/DB names).  That suppression could be relaxed
in the only-one-encoding mode, which would be a genuinue benefit.

			regards, tom lane