From: Tom Lane <tgl@sss.pgh.pa.us>
To: Artur Zakirov <zaartur@gmail.com>
cc: Peter Eisentraut <peter@eisentraut.org>,
        PostgreSQL mailing lists <pgsql-bugs@lists.postgresql.org>
Subject: Re: pg_dump crash on identity sequence with not loaded attributes
In-reply-to: 
 <CAKNkYnyet7hCwhhaaZE_7fBGCc9S6Xa9d4_Yy4YWqspdR9x4Sw@mail.gmail.com>
References: 
 <CAKNkYnwXFBf136=u9UqUxFUVagevLQJ=zGd5BsLhCsatDvQsKQ@mail.gmail.com>
 <4137493.1733781111@sss.pgh.pa.us>
 <CAKNkYnwTypJPsc258ZwVJ+n_Ti-8VxiQ6Gy4oMPmCA9fKo+KKg@mail.gmail.com>
 <66804.1733844592@sss.pgh.pa.us>
 <CAKNkYnwR0iLr2YeB9dkDXqh1vXvCyx9VyYTxw_mppCbkjqwP0A@mail.gmail.com>
 <84249.1733854089@sss.pgh.pa.us>
 <CAKNkYnz-uX2fk_V6bdp+KvFTCU8RtPG-k8MfMkgWRnM7nHoyQg@mail.gmail.com>
 <191373.1733865316@sss.pgh.pa.us>
 <CAKNkYnyet7hCwhhaaZE_7fBGCc9S6Xa9d4_Yy4YWqspdR9x4Sw@mail.gmail.com>
Comments: In-reply-to Artur Zakirov <zaartur@gmail.com>
	message dated "Wed, 11 Dec 2024 12:38:59 +0100"
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <482445.1733932129.1@sss.pgh.pa.us>
Date: Wed, 11 Dec 2024 10:48:49 -0500
Message-ID: <482446.1733932129@sss.pgh.pa.us>
Archived-At: 
 <https://www.postgresql.org/message-id/482446.1733932129%40sss.pgh.pa.us>
Precedence: bulk

Artur Zakirov <zaartur@gmail.com> writes:
> On Tue, 10 Dec 2024 at 22:15, Tom Lane <tgl@sss.pgh.pa.us> wrote:
>> The attached patch also gets rid of the dubious coding in
>> getPublicationNamespaces.  We might get push-back on that ignoring
>> schemas belonging to extensions, but if so I'd prefer to see the
>> behavior coded in a more transparent fashion.

> But other than that it seems the patch breaks dumps of mapping of a
> publication with the public namespace. The issue with the public
> namespace is mentioned in the original thread of the feature of adding
> schema mapping [1]. I couldn't find in the thread if it was addressed
> in the end. With the new patch the following mapping won't be mapped:
>     create publication pub_test for tables in schema public;

Ouch.  Okay, I agree that's probably bad, but then it raises the
question of why this filter exists at all.  It certainly seems
100% magic and undocumented that it does the "right" thing for
these cases.

In any case, that's mostly orthogonal to the problem with identity
sequences.  I now propose leaving out that change and just pushing
the change in getOwnedSeqs().  The publication-namespace business
should be discussed on its own thread.

			regards, tom lane